┊・storage

[pearl coral]

AIX hates all protocols lol...they love lvm disks

HP-UX was the same way

veritas filesystem ugh

[elfin igloo]

Just to follow this up, there is now an #1080541653036699739 section to ask questions 🙂

[little geode]

Has anyone here tried Ceph?

[vocal monolith]

Yes, it’s an easy deployment. Time needs to be spent optimising and configuring for your workloads. Getting the right performance out of it is kind of an art form.

[little geode]

What happens when you add a new, larger hard drive to an array? Lets say you have 6 2tb drives. And you add a 16tb drive. Would you get all that new space instantly?

@elfin igloo @vocal monolith

[cedar raptor]

You mean would you see an additional 16 TB of space?

no, it doesn't work that way. I don't think in any RAID type solution. even SHR, will just add in the 2 and leave the rest for later as not useable till you start to add more drives.

ONTAP - Will right size a drive if you're adding it to an existing raid group.
You can create a new raid group with large sizes. e.g. one rg of 2TB drives and another of 16TB drives.

[solid cedar]

Let's not jump straight to tagging people yet @little geode

[cedar raptor]

We're world wide here. both of those two are not US based either. they're probably sleeping at this point or close to it.

[vocal monolith]

It’s not a straightforward answer, depends on a number of factors and how you have OSD’s/weights configured. If you did have the larger drive configured to allow the use of all the space you would create an IO bottleneck as it would be getting more of the reads/writes as a percentage of the total. This can be controlled for reads but ultimately it wouldn’t end up great.

That’s why usually most systems will recommend uniformity of disk size and structure within a single logical array. The performance penalties are too large a trade off.

[obsidian mauve]

Hello .. Does anyone have template for Zabbix from NetApp FAS 2552-R6?

[sudden geode]

Hey, just asking if anyone knows about this: recently discovered on a users repertory in a UNIX qtree that some files were, at some point, renamed with always the same pattern. eg:

/home/toto/buro/:0300000
/home/toto/buro/:0300002
/home/toto/buro/:3NKDKQP

So far, it only happened for people working from Windows. Any idea?

[somber plaza]

Could be malware/ransomware?

[sudden geode]

It's very random to where it happened. No continuity at all and most important you can read the files without issues. I assume a ransomware would have encrypted it?

[somber plaza]

Typically. Could be an application writing some sort of temp files then? could try using less and see what pops out.

[sudden geode]

Looks like it's typical random average files, like a .docx. I managed to open one with Word and they have clearly been replaced, it's not new ones.
No idea when it happened, but we noticed this after an important move operation between 2 volumes, which we did with rsync. I want to say it's certainly related but then we observe the same patterns in the source and the destination 🤷‍♂️

So maybe it's not related at all, that's weird

I'm starting to exchange mails with our supplier who often kindly provide us with technical assistance but thank you for your time if nothing immediately strikes your mind about it, much appreciated!

[stable grove]

If it's still ongoing you could enable NAS auditing to find out what is going on: https://docs.netapp.com/us-en/ontap/nas-audit/index.html

https://docs.netapp.com/us-en/ontap/nas-audit/file-directory-access-events-audited-reference.html

[knotty shadow]

Hi, I am looking for this caddy but the only ones I can see online end with 0 instead of 1

https://www.piospartslap.de/?a=10167&lang=eng
https://www.bargainhardware.co.uk/netapp-111-00721-a0-sff-hs-caddy

Model: NAJ-1501

[knotty shadow]

Can anyone advise?

[cedar raptor]

which number are you looking at? the 111-00721?

AFAIK, the A1 is just a rev number. should still take a disk.

Maybe ask in #1062049107096633454 and see if a hardware person replies?

[little geode]

Could you guys look at this forum post and tell me what you think?
https://forums.lenovo.com/t5/Enterprise-Storage/The-difference-between-DM-and-DE-storage-devices/m-p/5206235

[patent flame]

I wanna give a to the core ONTAP/tools team, because this is quite literally one of the easiest upgrades I've ever done. 20 minutes ago I was on 9.8RC1.

• Download ONTAP image
• Upload image to SysMgr
• Click Button
• Watch YouTube videos

@twilit pendant @wise wave Bravo. Well done to the entire team.

9.8RC1 > 9.10.1P3 > 9.12.1P1

[native meadow]

And wasn't that always the case?
As easy as pie.

[patent flame]

I mean, to an extent. Before there was Tftp, manual cluster-failover mgmt, all kinds of steps. The upgrade itself was easy enough, but still a lot of manual steps that may or may not work. This is just next-level easy.

[uneven roost]

In all kindness, what I think of that thread, and your previous comments in here and on reddit, I think you have “analysis paralysis”. We have staked our company on ONTAP and Santricity. They’re both good. The Lenovo person has given you good advice on the differences. Between NetApp and Lenovo Santricity, the difference is who provides support and parts. Lenovo is a great partner and does a great job of both of those aspects. You may find it easier to deal with Lenovo commercially, but I think you might find a lot of benefit working with one of our partners one on one to help you architect your systems. You should start getting some prices since some of the numbers your mentioned in here are typically a bit lower than the investment point starts at for our gear. Some would be in there, but it depends on the capacity you want

As some high level thoughts on your queries - ontap is unlikely to be a cost effective choice for iscsi or nvmeof for backing a zfs pool. It would technically work, but what’s the point? Just use ontap to serve the data. Santricity might be a better option, but again, it does the data protection through raid/ddp so you’d use zfs for snapshots etc.

[vocal monolith]

This whole things needs a “Start with why” approach. What do you want to do? I see no benefits in digging so far into the weeds of specific solutions and technical terms that might have no benefit to the solution you are trying to build.

[little geode]

@uneven roost In all kindness, I dont need prices, I need to know about features but Lenovo and Netapp employees are contradicting one another about the features of Santricity. Obviously, this is causing frustration.

[stable grove]

What information are you missing? Both SANtricity and ONTAP protect against bitrot (as long as you keep the appliance running). There is absolutely no reason to fear corrupt data by bitrot.

[dark sphinx]

I can't speak for SANtricity as I don't work with that product enough, but ONTAP's file system is called WAFL. WAFL is perfect for SSDs because data is constantly overwritten to new free stripes.

(as new writes come in then the old stripes are freed)

[little geode]

@stable grove Ok thats what I needed to know. The Lenovo employee suggested otherwise and I was thinking, maybe their version of Santricity is somehow lacking in features.

[uneven roost]

I’m sorry this is still an area of concern for you. I thought I was able to help clarify this a month ago for you - #┊・storage message ?

[little geode]

@uneven roost You did at the time. And then I replied to the Lenovo employee with what you said, then THEY responded as though they were arguing against what you said. I believe they mentioned routine error checks (sounded to me like a Snapraid type of scheduling?) And this is why I brought the topic back up. I don't feel like going back and forth when the two teams should be communicating with each other. (Lenovo employees are undoubtedly the lessor experienced of the two). Maybe you should login to their forum and discuss the issue so others can have the answer via google searches?

[velvet scaffold]

Hey guys,
Anyone know how Ontap snapmirror select which snapshot it will transfer to the backup?
Lets say you have 7 daily snapshots on primary on 14 on the secondary and the snapmirror runs daily.

How does it know that it only have to transfer the new daily snapshot? Is there some hidden ID where it know that it already transfered the 6 other snapshots?
And if you have the answer, do you know where I can find it in Netapps Official documentation?

[uneven roost]

destination system "pulls" from primary, and retains based on policy for retention. So the destination knows what it should keep, and primary knows what it should keep, and they don't have to be the same. Page 27 of this is the best guide - https://www.netapp.com/media/17229-tr4015.pdf?v=127202175503P - "Protection Policies"

[uneven roost]

thanks for your feedback. I will pass it on to our technical liason for Lenovo for them to clarify

[velvet scaffold]

I'm onboard with that. But I wonder how does the destination know what to pull and what has been pulled. Given that all the snapshots have the correct labels.

[uneven roost]

the destination knows what it has and what it needs by policy and pulls more copies on that basis and deletes local copies if it has more than it needs

[tawny basalt]

The labels, as far as I recall are mostly used with a vaulting relationship.

You can look at the different snapmirror policies and see what they do. For dr, I generally use MirrorAllSnaoshots

snapmirror policy show

[thorny maple]

TMAC is correct. Previously Snapmirror used DP and SnapVault used XDP, but, since (I think) OnTap 9.6, they both use XDP by default and you control what gets transferred based on the Snapmirror policies.

If you look at the snapshots on the source volume, they should have a Snapmirror label:
snapshot show -volume vol1 -fields snapmirror-label

The snapmirror policy that is used to transfer the data to your secondary location uses a policy:
snapmirror show -destination-path vserver:vol_name -fields policy

Then that policy dictates which snapshots will be transferred and how many will be kept:
::> snapmirror policy show -policy policy_name_snapvault_0h_30d_4w
Vserver Policy Policy Number Transfer
Name Name Type Of Rules Tries Priority Comment

---

cluster01 policy_name_snapvault_0h_30d_4w vault 2 8 normal -
SnapMirror Label: daily Keep: 30
weekly 4
Total Keep: 34

For the policy above, we are transferring daily and weekly snapshots, then keeping the last 30 dailies and the last 4 weeklies.

[sand condor]

@thorny maple ONTAP 9.3 made XDP the default policy. 9.6 eliminated TDP support until it was reinstated in a 9.6 P-release.

[little geode]

is it normal that Lenovo just doubled their san prices?

[uneven roost]

That’s not the general trend we are seeing across the industry. To say pricing has doubled would suggest some difference in quoted product or a quoting discrepancy or problem either initially or recently

[little geode]

I doubt they were having a genuine 50% off sale across their entire inventory.

@uneven roost

[uneven roost]

Yes exactly. I also doubt they’ve just doubled their prices. In the end their pricing is very much their own business and not something we can comment on specifically, all I can say is that generally I would guess what I said earlier - the quotes are either for different things or there was a mistake with one of them.

[little geode]

But its across the board. For all their datacenter products. Servers and sans and workstations. @uneven roost

[uneven roost]

Interesting.

Maybe some other members have seen similar things.. I actually don’t even know how much our stuff costs, let alone others

[little geode]

Prices are now back to normal. Very weird.

[cedar raptor]

where are you looking at pricing. only List is ever published. and i think even those lists are limited to express packs.

[little geode]

lenovo.com

[little geode]

Does Netapp E series do checksum verification on reads?

[elfin igloo]

Please post your question in the #1080541653036699739 section. Thankyou.

[patent flame]

Gettin a little bit o' boogie outta the 250 this week!

[edgy marsh]

Any tricks for the unreliable sensors on disk drawers on DS460C shelves (and equivalent on E-series and StorageGrid)?
We have some problems with them sometimes. I was just onsite at a customer to reseat a drawer, and the "Drawer open or removed" message came back after a few hours.

[drowsy dust]

Is there a decent way to determine from the CLI whether the volume you're looking at is protected via snapmirror?

I've got like ... 2000 vols and aside from going 1 by 1.
Doing loose counts, I'm definitely missing some, but id rather not have to export them to excel and compare there.

[cedar raptor]

snapmirror list-destinations?

[drowsy dust]

Shows me all that exist, not ones that don't have destinations.

I know system manager can filter on this, but I can't export that into anything usable

[drowsy dust]

Eh just gonna do this in PowerShell

[edgy marsh]

Could you get something from looking at what volumes have busy snapshots? I don't remember if the "owner" field actually can tell if a snapshot is busy because of snapmirror

[drowsy dust]

I'll investigate that route tomorrow AM.
I forgot about the owner field until earlier today when cleaning up screwy snapmirror snapshots from a month ago

[uneven roost]

make sure IOM firmware is totally up to date, but if it's the same drawer that keeps faulting, schedule downtime to clean the optical sensors

[edgy marsh]

The last shelf with the problem is a StorageGrid SG6060, so it's an E-series E2800. Whend reseating the drawer, I noticed the fault LED going off when I slooowly pulled the drawer out the first mm. Then it came back. We have also cleaned the sensor on another shelf, and that did solve the problem.

[edgy marsh]

snap list -owners snapmirror
Does this return what you want, @drowsy dust ?

[drowsy dust]

Negative!

[tired radish]

Anyone know how long it takes for the hardware visualization in ONTAP gui to update and reflect node name changes?

[patent flame]

Should be pretty instant. Went through this recently myself.

[tired radish]

been a few hours

[cedar raptor]

Try incogneto mode?

i'd also maybe check the cluster ring and see if that's all good

[tired radish]

same in incognito, cluster ring seem sfine

wonder if its confused since I swapped the node names to reflect tha correct A and B sides when viewed from behind

[cedar raptor]

shouldn't matter. the names are just for us humans.

[tired radish]

We have a headswap on this system tonight, hopefully this wont be the cause of issues

[cedar raptor]

what ontap version you on?

[tired radish]

9.11.1P7

[cedar raptor]

thx. not seeing any KBs or BURTs related to that.

anything in EMS? other wise, a support case might be in order.

oh. check the vreport -

set d; debug vreport show

[tired radish]

running the command, seems to take awhile

[cedar raptor]

that is also odd. usually pretty quick to return even when it finds errors.

[tired radish]

seems like a dns issue, had some weird routes in the cluster vserver, cleared that and its fast now

set d; debug vreport show

This table is currently empty.

Info: WAFL and VLDB volume/aggregate records are consistent.

its nagging in my brain, wonder if I should swap the names back before the headswap

[cedar raptor]

it is odd for sure. I'd open a case and have support give it a once over before PS comes on site for the swap. just to avoid any delays they might catch when starting the swap.

[tired radish]

thanks for the help Mike, appreciate it.
I did another rename and it seem to take this time around

[cedar raptor]

no prob. guess it just needed a poke? 🙂

[patent flame]

Some light reading material for you storage aficionados. https://www.netapp.com/blog/ontap-reaches-171-gpudirect-storage/

[cedar raptor]

https://tenor.com/view/rick-bobby-will-ferrell-talladega-nights-fast-indy500-gif-4552859

[tawny basalt]

Set advanced
System controller config show

Look for the chassis/loader slot ( it will be either A or B)

Viewing from the rear,
The left/upper slot is A
The right/lower slot is B

[buoyant apex]

Hello SE friends, what is the best way to get some projections about the dedupibility of a data set which does not currently reside on dedupe-capable storage? The data can't leave the site, so no ability to copy it to a NetApp to find out that way.

I know the space savings estimation tool (SSET) used to be a thing, but all links refer to XCP now, and I'm not sure that dedupe/compressibility actually a function of XCP's File Analytics.

[ebon drum]

hello

[drowsy dust]

Your Sales team should be able to help with this an like-like projections on storage

[little geode]

I was sent this via PM from a storage expert who prefers zfs over all other options and I would like feedback from this room. I wont reveal who it was. And my main concern is that netapp no longer does all of the stuff he mentions.

There are a few that are actually good [enterprise sans]… but they tend to be the most expensive options. For them to be good they have to do checksumming and checksum verification on reads; produce reports about corrections made/scrubs/detected errors, provide tools to help detect and monitor mismatches between stored/accessed data and dataset geometry (ie. tools to look for things like… write amplification, misaligned data boundaries, synchronous write performance degredation, stats on i/o latencies and throughput on a per drive/shelf/cabinet level, etc. these were all bog-standard capabilities on 15 year old netapp boxes, for example, and continue to be present on some storage systems).

[drowsy dust]

Personal opinion: hard core ZFS people are a chore.

That aside, those absolutely used to be reports you could get on 7mode systems, but the modernization of storage abstracts that grain of sand counting from the consumer/end user.

You can possibly get some of that from the logs, or node level commands

[little geode]

What about " checksumming and checksum verification on reads"

[drowsy dust]

Meaning running a checksum on each individual read?

[little geode]

Yes

[drowsy dust]

https://kb.netapp.com/onprem/ontap/hardware/What_are_the_important_changes_to_RAID_scrub_in_Data_ONTAP_8.3.x_or_later

For scrub details.

I don't know what to tell you on read checksums.

[little geode]

So what you are saying is, the zfs people could be right?

[drowsy dust]

I'm not Netapp staff.
But I don't imagine that there isn't a function for read checking.

My personal opinion is ZFS people will find any reason to smear everything not zfs

There's also been a rash of "product Y doesn't do ANY checks recently" which is wild to me.

[drowsy dust]

Because I also, can't let stuff go, the last time you asked @little geode I think you got this answer https://docs.netapp.com/us-en/e-series-santricity/sm-storage/data-integrity-and-data-security-for-volumes.html from @elfin igloo ((sorry for tag, but you deserve credit))

[patent flame]

I’m gonna put ”ZFS People are a Chore” on a tshirt. 🤣

[cedar raptor]

I want - SAN people always ride single file to hide their numbers.
on a shirt.

[uneven roost]

Eric Davis's "e5a and e5b really do exist” shirt is awesome

[uneven roost]

@little geode Our systems specialise in data integrity, it's a core product design feature. For everything you've asked about, the answer has been yes, and there's any number of other reliability and survivability considerations we make in designing systems that aren't even on your radar. I think you need to start looking up the stack at management, support and pricing.

and work out if ZFS or ONTAP/E-Series is what you need

[little geode]

Why do you guys get defensive when I ask for details? Also the linked pdfs and info pages dont mention any of those features.

[uneven roost]

because we are an appliance vendor - our best fit for customers are people who ask these questions and are satisified, or ask more questions. The same question over and over again is a suggestion this might not be the product for you and we'd like to help you come to a decision, even if it isn't us.

because in the end we're storage nerds, and we want people interested in storage to take the jump in whatever direction they'd like

this was the reason for my question on if you had pricing already - you may find we're in a different level to where you're looking at investing for your company

[little geode]

Or... you arent answering my questions about features?

[uneven roost]

you've asked if we do data integrity checking in about 5 different ways by my count. The answer is yes, yes we check data integrity

we aren't parnasus 🤣

[little geode]

Data integrity is a marketing word. You know that. 🙂 The DETAILS of what happens and how it compare to zfs, this is what i have wanted to know

compares*

[uneven roost]

in 2022 we made USD$6.3B in sales based on the quality of our products. ZFS didn't.

sooo.. our customers are generally quite happy with data integrity

[solid cedar]

@little geode
There’s a depth of conversation here that seems to have reached the outer scope of what we can help you with here in Discord. If you’re comfortable with it, please DM me your corporate contact details so I can help get a rep and/or SE connected with you.

[stark mica]

Hi All, I have just joined discord. Am I allowed to ask questions about NetApp StorageGRID here?

[uneven roost]

Sure can - check out #1063547220009439353

[stark mica]

Thank you for mentioning the storagegrid tag. I went there and searched a keyword, but it did not look my keyword was discussed earlier.

[elfin igloo]

You can submit new questions there also.

That would be the best place to submit any StorageGrid related queries.

Hey @stark mica seems there is some confusion. You can head over to the #1063547220009439353 channel and submit your question there.

Thanks!

[stark mica]

Ok, I will do it. Thank you.

[cedar raptor]

I loved seeing that shirt around insight.

time to put the Cricut to use and make some shirts for Converge and INSIGHT.

[drowsy dust]

Only if I can get one lol

[drowsy dust]

I don't represent Netapp and never have, I linked and got snippy, my opinion was stated clearly as to why lol

[patent flame]

📢 NetApp takes #1 spot in EDA Benchmark 📢

Read the Full article ➡️ https://www.netapp.com/blog/spec-bechmark-eda-solutions-top-spot/
SPEC Results ➡️ https://www.spec.org/storage2020/results/eda_blended.html

NetApp posted the #1 result for the SPECstorage Solution 2020 EDA_Blended benchmark, which focuses on simulating the EDA process from end to end. Our results show that the NetApp storage environment delivers:

Predictable, consistent high performance, which results in more concurrent job sets than any other vendor
Better scaling, with continued low latency as demands increase from more EDA workloads

[dark sphinx]

That's awesome.

[dense wave]

hi, I want to buy a supported ds4243 device and connect it to my hba lsi 9400-8i , will this build support 18 -22 tb sata hdd?

[drowsy dust]

Home lab setup?

If so, it's whatever the controller supports at that rate.
I'd look at upgrading the IOMS from the IOM3 to IOM6 if you can

[dense wave]

yes

[dense wave]

I will have enough speed on ion3, I want to learn about the support of larger capacity sata disks. because I had a supermicro server. and his backplane did not support disks larger than 10tb, I had to sell it.

[drowsy dust]

Yeah these shelves act as a pass thru if I recall.

[patent flame]

You said "supported" and I wanna be clear that this is not a supported configuration by NetApp. Will the DS4243/4246 shelves run as JBODs? Absolutely, some of the best out there, but connecting shelves to anything but NetApp controllers isn't supported.

That out of the way, yes, any of the LSI 9200/9300/9400 HBA's will work. Be wary of using RAID controllers instead of vanilla HBAs, as that can lead to some issues. But most of them can also be flipped to "hbamode=true/on" to just connect JBODs.

[dense wave]

That's right, I'll use a direct connection, without a raid. there are videos on YouTube, but 16tb disks were used there, and I wonder if 20-22 tb and future 30tb disks will work

[patent flame]

No reason they shouldn't. I would near-insist that you take the IOM3 modules in the DS4243 and swap those with IOM6 modules. This is the difference between 4243 and 4246. IOM3 is 3Gbps and IOM6 is 6Gbps SAS. Most current SATA drives are running 6Gbps, if not 12Gbps. Also, you're going to want to insure you're getting 7200rpm drives, not 5400rpm drives. You're also going to need some QSFP to whatever-miniSAS-connection-your-HBA-has transition cables

[drowsy dust]

Yeah those rebuild times will be murderous on 3gb interconnects.

[dense wave]

there is no way to order it yet. I will be used through a transition board.

mini sas to QSFP

[patent flame]

Order what?

[dense wave]

I don't need a lot of speed, I will only use it as a file storage

[dense wave]

there is no money for iom6 ((

[patent flame]

eBay, they're $25 a piece

[dense wave]

there is no delivery to my country. in stock in my country there is also a cost of 450 dollars

[drowsy dust]

Yikes.
Plan accordingly for slow rebuild times then.

[patent flame]

Fair enough. I hate that for you. The jump from 3Gbps to 6Gbps is monumental.

[thorny maple]

I'm curious about the backstory behind that shirt.

[uneven roost]

There’s an notification message that shows during boot up on some platforms/ontap versions that says something “e5a and e5b do not exist”.. in some cases it’s platforms with no possibility of an e5, but in others there /may/ be an e5a/b - but the message is triggered unless the system is setup as an IP metrocluster

[cedar raptor]

Yeah. First time I saw that in the field was like a 9.3? Deploy and scratched my head a bit.

[light sigil]

Noob question but I haven't found a straight forward answer online yet... I purchased a DS4486 on ebay and filled it up with 48 drives. Does it support a hardware raid? I've bought a few sas cards and only been able to get it working in HBA mode. The Raid Cards I got don't seem to let you add drives from a enclosure to an array. Not sure if my best bet is to return all this stuff or what? Seems hard to believe a 8 port raid card could suddenly support 48 drives. So I must be missing something fundamental.

[cedar raptor]

those are all JBODs, there's no intelgance / CPU / OS etc. to them.

and they are ment to connect to systems running software RAID. (ONTAP)

[light sigil]

That is what I was afraid of. I guess I need to look into software raids.

[cedar raptor]

there are people that do mod it. though i've not heard of someone using a 4486. just due to the nature of the caddies.

[light sigil]

Is there a good resource for comparing the hardware raid to software raid. Not sure I trust the results I'm getting in Googles. Some bias either direction.

[patent flame]

If you just wanna do unRaid or something of the sort, you don’t even need a RAID controller. A simple HBA to present the disks will get it done. Stick to LSI 9200/9300 stuff.

[light sigil]

Yeah I have one of those. I just wanted to raid it. It seems like that isn't a thing? I haven't considering unraid since I'm running windows server OS on the machine with the card in it.

[drowsy dust]

Get a quality HBA on the approved vendor list for whatever homebrew solution you're using and whammo you're in biz

[patent flame]

Grab an HP P card, like a P410 or P822 with some external 6Gbps connections. That’ll get you some hardware raid. They’re ~$25 on eBay.

[light sigil]

Hmm that p822 sounds promising

[drowsy dust]

With those cards, if you value your data don't cheap out and not get the battery, it's extra peace of mind.

[tired radish]

Curious with 111-04739 - 2p 100GbE RoCE QSFP28 modules.
Is the command to change mode between storage and networking an adapter wide setting or per port?
For example, could we have one port storage and one network on the same adapter?
The documentation makes it seem like it but wondering if changing one port forces the other to be the same

[drowsy dust]

System and environment?

In ye olden days, the CNA ports could be swapped between FC and Ethernet with a few commands and sometimes a reboot.

[tired radish]

Its a A900 with ontap 9.11.1P7
I think some CNA cards force port changes in pairs due to asic configurations I think.
Wonder if this card is the same

[cedar raptor]

Can you post in #1062049107096633454 ?

[tired radish]

sure thing

[charred tapir]

Hie Guys, Can anybody please help me with Health checkup report/ Assessment report of NetApp storage for FAS, AFF & E-series?
Like exactly what we do or what we check in Health checkup report/ Assessment report of NetApp storage for FAS, AFF & E-series?

[uneven roost]

Hi @charred tapir, are you a NetApp employee?

I think it’s ok to share.. we have an internal database of thousand of configuration/status risks, which is used by ActiveIQ, ActiveIQ unified manager and a number of other tools

So.. we check a lot of stuff.

[minor acorn]

As an EDA customer, this is no surprise. I have run a large compute farm on vendor's storage as well, but won't switch away from NetApp now. And it's not just about performance either.

[native meadow]

Hi.

I got a FAS2552 system bornt as a single node on 9.8P17 with ADP.
Can I add a second node to It?

I saw that the partitions that would be assigned to the second node root aggregate are available as spares to the first node.

Is there public documentation for guidance?

[uneven roost]

Should work. No docs afaik.

I’ll assume for the purposes of license compliance that you already own the other node and it has the same licenses

I would say install the second node, boot to maint mode, use sysconfig to find the system ID, assign the appropriate number of boot partitions to it, then NetBoot the new node if it wasn’t already running 9.8P17 and perform boot device replacement procedure.. then just normal boot and it should start the node install process? There might be a few more steps along the way

Start a thread in #1062049169520476220 when you hit them 🙂

[native meadow]

Thanks, @uneven roost.
The node was netboot and is already running 9.8P17.

I will check tomorrow at the end of the day and let you know.

[atomic jay]

hi

does anyone knows if we can move netapp snapshots to s3 compatible object storage?

[cedar raptor]

Fabricpool / Cloud Tiering. https://docs.netapp.com/us-en/ontap/cloud/fabricpool-concept.html

there is a snapshot only policy

[atomic jay]

tks @cedar raptor , but if this Object Storage is not Storage Grid? What if we if we've ECS or HCP.....is it still possible? probably not using fabric pool, but any other option?

[cedar raptor]

There's a supported list here - https://www.netapp.com/pdf.html?item=/media/17239-tr4598pdf.pdf. page 39

[stoic jolt]

How are people going with all their ONTAP upgrades for SU530 ? Any interesting stories ? What is the oldest version of ONTAP you have to upgrade from.. ours is 9.3P4.. im tossing up which way to that.. do you do 9.3P21 -> double hope to 9.7P22 or do you just go 9.3P4 -> 9.5P19 -> 9.7P22 ... for those still with 25xx are you stopping at 9.7 or going 9.8 with new system manager?

[patent flame]

My oldest is a 2650 and I got it up from 9.8 to 9.11.1P12 I think (whatever the latest P release is)

If you've got older than 9.8 I believe you have to two-hop it to 9.10.1 >> 9.12.1P2

[stoic jolt]

the biggest stoppers of ONTAP upgrades now are 26xx hardware (9.11.1P*), CN1610 switches (9.11.1P*) and 25xx (9.8P*), i understand all these are towards EOL but there is still a heap around just ticking away in DC's

[patent flame]

I hope to have some news for you at or before Insight to address that. I'm pushing a boulder up a mountain though, so bear with me.

[minor acorn]

I've been checking my client auth methods and all are using kerberos so we don't seem to be affected at all. We'll upgrade ONTAP anyway but that's because we want the updates for other reasons.

[tired radish]

I see ontap 9.11.1P9 just out - don't see patch notes

[stoic jolt]

from my looking into it, via IP is typically NTLM, via name/UNC is kerberos but also depends on config on their DC's.. so its really a lucky dip.. seems all our customers are mixed

[patent flame]

Here's the document, and in the Solution pane if you scroll down, it lists which P-patches update to fix the Netlogon RPC issue.

https://kb.netapp.com/Support_Bulletins/Customer_Bulletins/SU530

[chilly axle]

Wasn't SU530 addressed in P8?

[balmy dune]

I am not sure if this is the right channel to ask that, tell me if not... Does anyone have done third party storage transfer to NETAPP storage? I'm in the process of replacing our 3PAR storage with a new NETAPP environment. Is there a tool available that allows for direct data transfer between 3PAR and NETAPP? I have a significant amount of data (TBs) to migrate, and I'm looking for the most efficient solution.

and Hi everyone 🙂

[patent flame]

Is it block or file?

Since you said 3Par I’m assuming block

[balmy dune]

Yes, it is block.

[cedar raptor]

What’s the workload? VMware is usually just storage vmotion

Bare metal luns. FLI is worth a look.

https://docs.netapp.com/us-en/ontap-fli/san-migration/concept_foreign_lun_import_overview.html

[balmy dune]

Bare metal luns and a lot of data. Yes, I was looking into FLI. thanks for pointing it out.

[cedar raptor]

What platform?

AFFxxx?

The exception to FLI is the a250 and fas500f as it doesn’t have an initiator card.

[balmy dune]

FAS2720

[cedar raptor]

The cna ports can toggle between 👍🏼

[stoic jolt]

Note:
This release contains just one fix for a specific customer need. Bug ID: 1550611 - AWS-FSx: add retry for IMDSV2 token.
There is no other change from the 9.11.1P8 release.

[stoic jolt]

Anyone know when SMBC is going to support NVMe over TCP, given the re-launch of the ASA is this a 9.13.1 thing ? we have some customers with A250 and A400 with NVMe drives and want to move over to NVMe over TCP for VMware but they need to use SMBC in a VMware stretch cluster for HA ?

[kindred pendant]

Hello, is there any easy way to obtain qtree overcommitted report by its volume's maxsize? through OCUM or through API?

[elfin igloo]

Hey Greg, you may get a better response by dropping your question into the #1062049169520476220 or #1062049107096633454 sections.

[elfin igloo]

Hi Senthil, thanks for jumping into the NetApp Discord and asking your question! Suggest dropping your question in our #1062049169520476220 support/questions area or even the #1063542514780475493 if you want to get into the API side of things.

[obsidian mauve]

it's there a way to get the "Powers On" from a single hard drive on FAS-2552?

[past sun]

**:bar_chart: Do you segment your SMB/CIFS network from core hosts network? **

🇦 Yes, we use VLANs
🇧 Yes, we use non-routable subnets on discrete switches
🇨 No, we use the same subnet for AD and Storage traffic
🇩 We don't segment at all
🇪 Other (please describe below)

[patent flame]

All, I'm doing some discovery for SysMgr enhancements on SVM creation workflows. Please let me know how you prefer to lay out your networks. I'd like to feed this data back to the SysMgr team.

[stoic jolt]

we have many customers and all have different setups/requirements most have dedicated vlans.. one thing which is common is we don't ever use those ONTAP system manager workflows since 9.8 is become so annoying. Its command line

[patent flame]

That’s understood. I promise it’s made great strides since 9.8 and we’d love the feedback on the current iterations. Intimately aware of the issues people have with 9.8, but even the differences between 9.11 and 9.12 are massive.

[elfin igloo]

Hi @obsidian mauve suggest dropping your question over at the #1062049169520476220 or #1062049107096633454 areas. Thanks!

[stoic jolt]

Agreed. They are getting a bit better. but the thing is a lot of the newer wizards want you to do everything in one shot.. often im not in a position to do everything at once (waiting for other teams). back in 9.7 you were able to skip steps and it would still work..Eg creating a CIFS SVM wants me to connect to AD right away. 1/2 the time i don't have DC credentials and i need to screen share with a Domain admin to get it done.. In the mean time if i create it with CLI i can create the SVM, Create Volumes, Create network interfaces.... and then when the domain admin is free i get him to whack the creds in.. BTW this is on 9.11.1 i have not tried 9.12.1 yet

[stable grove]

Absolutely agree on this. Not being able to create volumes in System Manager if no running CIFS server exists is stupid.

[drowsy dust]

I go full tinfoil.

All management interfaced are in a VRF only accessable via specific hosts.
Web UI is behind saml
Ssh is not doing 2fa, but the jump host itself is.
So if you want sysmgr it's painful.

• X11 fordwarding with 2fa
• Hit system manager with saml 2fa
  All client or protocol facing traffic are on separate storage vlans depending on the segment of the organization they belong to.
• Research
• Clinical
• Dmz
• Etc.

I go beyond the extra mile.
My colleagues hate me lol

[dusk lava]

@patent flame is the tl;dr of SU530 basically "If you haven't patched ONTAP by June 13, you're going to have problems with unpatched Windows hosts"?

[patent flame]

Only if you're using basic NTLM/Netlogon authentication. Kerberos and FIPS are not affected by the CVE.

TL;DR - It's being rolled out via Windows Updates to all domain controllers that will change the RPC calls that Netlogon uses to authenticate. So, once your DC's are updated, if your ONTAP isn't updated to support those new RPC calls for auth, you're gonna have a bad day.

We've made P-patches for all versions supported that are available now, all the way back to ONTAP 9.7...

9.7P22 (published April 11, 2023)
**9.8P18 **(published April 19, 2023)
9.9.1P15 (published April 7, 2023)
**9.10.1P12 **(published April 25, 2023)
**9.11.1P8 **(published April 28, 2023)
**9.12.1P2 **(published April 10, 2023)
**9.13.0P1 **(Published April 12, 2023 as a Cloud Volumes ONTAP specific release)
9.13.1RC1 (published May 4, 2023)

Customers will need to upgrade to one of the above listed releases in order to be able to obtain the enhancement in ONTAP that will allow the continued use of NTLM/Netlogon authentication past Microsoft's final Enforcement phase.

(Yes, there are some finer details/points about the 4 phases of the roll out, but do yourself a favor and just upgrade ONTAP to the latest P-patch of your release so that you're covered before the 4th and final "Enforcement" phase lands)

[thorny maple]

if you run a "cifs session show -auth-mechanism !Kerberos" on your OnTap cluster, it will show you any clients that are not using Kerberos that will likely be affected if you don't upgrade. I thought we were all Kerberos, but because of this discovered there are some Samba clients and Macs that are still using NTLM.

[dusk lava]

It could be worse, Jim.. they could still be using SMBv1 😮

[drowsy dust]

Listen, I've got like 2 apps that still do it.
They're on their own svm with nothing else, so they can be bad actors in the corner.

[elfin igloo]

In the "awful" corner. 😄

[thorny maple]

Trust me, we just got them off of that crap in the last year or so. Still had some old Xerox printers that required it for scan to file.

Can anyone recommend any good software or techniques for getting data off of an external USB drive that is no longer being recognized by Windows? My brother has one of those Seagate backup drives that is no longer being recognized and apparently he doesn't have the data anywhere else. I had him run through the troubleshooting steps from their web site, but no luck. Any other ideas?

https://www.seagate.com/support/kb/usb-external-troubleshooter-003581en/

[uneven roost]

Call Kroll

There’s often SATA access points on the controller board, but it’s not DIY

Assuming photorec doesn’t work

[solid cedar]

If the drive can simply be disconnected from the enclosure, I’d mount it directly in Windows. I’ve done it with WD external drives before in a similar situation.

[drowsy dust]

Shucking was my first thought if it's a single drive vs dual

Unless it's some abomination of raid 1 in there

[solid cedar]

I've got an old WD RAID-0 / dual drive enclosure that supposedly can be flipped to RAID-1 (somehow). It's 13 years old now. If it dies, it dies.

[thorny maple]

This is just a single 5 Tb USB connected external drive. I have a similar one, like this: https://www.seagate.com/content/dam/seagate/migrated-assets/www-content/product-content/expansion-fam/expansion-external/en-us/docs/expansion-desktop-ds1763-5-1404us.pdf

I will try to remove it from the enclosure, but there aren't any screws or anything in the enclosure. Looks like it was all glued shut. Definitely going to destroy the enclosure, but at this point, who cares?

[patent flame]

Heat gun is your friend

[prisma dune]

Do you guys know if ActiveIQ can be integrated with opsgenie? I searched and there is no mention of it . Can someone shower some info please

[uneven roost]

activeIQ unified manager can send events to an external destination via REST-API. I don't know of a specific integration with that, but it's worth trying

[prisma dune]

Thank you will look in to that.

[waxen frost]

Hi @uneven roost do you eventually have an example for that?

[uneven roost]

Not off hand.. I believe there’s some in the documentation

[dark sphinx]

See if Spinrite will work on a system with BIOS that recognizes USB drives.

[mossy ermine]

Depending on what exactly is broken, this has helped me in more than 1 situation: remove the drive from the enclosure, find out the exact model, buy second drive of same model, swap the PCBs

[solid cedar]

TIL that Steve Gibson at GRC publishes a weekly security blog, and that there's a patch for SpinRite that resolves an issue with drives >549GB.
https://forums.grc.com/threads/a-patch-for-a-spinrite-v6-0-overflow.1232/

The patch was just published last week!

[dark sphinx]

Yeah. 🙂

[pastel jolt]

Hello does anybody knows if netapp has an article with the IOPS expected from a storage array AFF 700?

[mossy ermine]

I doubt that a 20+ year old program that just uses regular DOS/BIOS access to a drive can recover anything more than what eg. dd or ddrescue can 🤷

[uneven roost]

We don’t publish or establish maximums - iops is a crude measurement of a storage system performance. It is heavily dependant on back end storage (what type of disks, how many) as well as cachability of data access patterns

[cedar raptor]

one of my fav bits on Wikipedia - https://en.wikipedia.org/wiki/IOPS

"Absent simultaneous specifications of response-time and workload, IOPS are essentially meaningless."

Throughput is (generally) your better metric.

[mossy ermine]

throughput is equally meaningless because otherwise everyone would be using EF-series instead of FAS/AFF 😉

[cedar raptor]

flexability, scaling, protocal support, multi-protocal support?
EF does one thing really fast.

[minor acorn]

IOPS are meaningless. I measured a 24-node Isilon cluster with SAS drives doing 1M IOPS. I've got a 10-node AFF800 array doing 1.5M IOPS. But those aren't reads and writes - they're metadata which may or may not be important to you. If I tried to do 1.5M writes per second, I might not be happy . Throughput is meaningless. I have tipped over an AFF cluster writing at 1GB/sec but it depends on HOW those writes take place. How your specific application responds under load is useful and measuring that ranges from HARD to VERY, VERY HARD.

[crystal solar]

IOPS aren't "meaningless"... you just need to be more specific about what type of IO ... it's nuanced. Obviously when buying a new system, there are sizing concerns.

[drowsy dust]

Is simulating Ontap via the sim in the tool chest still a "beat effort" type ordeal?
I've got a mighty need to test some changes before I can roll it out into prod and have no way to implement it without this thing (networking modifications and protocol functionality) and this finicky beast is refusing to boot past a point

[eager hatch]

I think ONTAP Select is the way to go for test systems, a bit overkill usually since it requires a license and deployment server but I've never gotten the ONTAP Simulator to work as a relatively permanent setup. I think it fills up the root partition or something and then just dies.
I think the answer to your question is "community supported".

[drowsy dust]

Yeah ive deployed it a few times as a point test.
"I need to test VRF changes for management and cifs"

[dark sphinx]

https://kb.netapp.com/onprem/ontap/Performance/What_are_IOPs%3F

[kindred pendant]

Is this a BUG? '/api/private/cli/snapmirror' won't return schedule key when using any filter.

[elfin igloo]

Consider posting your question in #1062049169520476220 or #1063542514780475493 for help.

[light sigil]

So I picked up a P822 but it seems it gets stuck on trying to boot to the option rom. Unfortunately the motherboard doesn't have a way to disable the option rom booting from the pci card.

Any other options?

[light sigil]

It seems the DS4486 has a interposer inside. This is what the raid card is seeing and not the raw drives. So I think that is what it won't let me raid it. Some Marvell Luigi V2.

[drowsy dust]

4486's are the ones with dual drive per bay, correct?

[patent flame]

Yes, correct. I would certainly prefer a 4246, but I can understand the desire for double-disk drawers and capacity in the same RU

[uneven roost]

yeah, afaik I haven't seen people have success for off-label use of DS4486s

not saying it's not possible, just that they use DS4246s more often

[mossy ermine]

4246's are also much much more common than 4486's at least here in Europe. And for off-label use, you'd have to make extra sure that your RAID group doesn't span over both disks in a carrier (and the disks in the carrier are exported as two LUNs of a single device, that might also complicate things during setup)

[light sigil]

Looks like I could have done a software raid with the 4486. I had a hard time finding a hardware raid controller that worked with that interposer.

Here is what it said for the 4486. Trying the 4246 now instead...

[mossy ermine]

do you have a high res screenshot of the interposer card? where the chip is clearly visible?

[light sigil]

Nope swapped it out for the 4246, but can search for X480A-R6 model and a bunch of them come up.

I did look up the chip to be Marvell 88SF9210.

But now my LSI card is throwing errors with the 4246.

[patent flame]

Are you trying to run hardware raid via the controller? Or are you trying to load up TrueNAS or unRAID or some other software solution?

[light sigil]

Hardware Raid via controller

[drowsy dust]

Honestly.
I wouldn't hardware raid

[light sigil]

I tried a perc h810 got the same message. So will just ignore it for now. Maybe dell management tool won't throw those.

[mossy ermine]

these are just messages from the SES enclosure. They should not affect access to the disks

[light sigil]

Yeah, but not sure if I should be worried about that or not.

[stoic jolt]

does anyone know if NetApp ONTAP works with IBM-I ? either 7.3 or 7.4

[uneven roost]

like IBM i-series Power?

Right now we have support with ONTAP 9.12.1 for up to AIX 7.3 TL1 SP1 on IBM Power

FC or iSCSI. please see IMT for more details https://imt.netapp.com/matrix/#playground

[mossy ermine]

IBM i, that's the AS/400, right? If so then no, it's not supported AFAIK

I think they use FICON on the IBM i

[stoic jolt]

It looks like you can use NetApp to mounts FC LUNS into VIOS and then represent them as 520 byte virtual SCSI (this is 100% possible, yet painful) , In the end the customer only needs a small amount 20TB so i might just stick with cheap and cheerful IBM flash disk as its all in the eco system and just works

[patent flame]

It's just being fussy that it can't talk to the fans and power supplies, or can't see that they're running. It shouldn't have any issue tracking the disks themselves

[light sigil]

Anyone know if getting a second cable for the DS4246 would make it transfer files faster to/from?

[uneven roost]

Probably not. There’s 4x6Gbit anyway, and 24 SAS disks would have a hard time saturating that (24x160MB/sec on a good day)

[mossy ermine]

on larger stacks it might be possible, but I agree that for a single shelf, all disks would need to essentially be in sequential-read for a second cable to have any effect

[stable grove]

ONTAP 9.13.1 is GA!

[patent flame]

ONTAP 9.13 Release Notes: https://library.netapp.com/ecm/ecm_download_file/ECMLP2492508

[patent flame]

Dear XCP, how do I love thee? Let me count the ways... my latest 50TB consolidation data move.

Total Time : 2d13h
STATUS : PASSED```

[minor acorn]

Let me count the ways - we did a huge migration a few years ago (from another vendor to NetApp). We moved all of the stuff that could be offlined first, but had a maintenance weekend for the active file systems. During a long weekend with all clients down, we migrated 1PB of data, some of which was seeing a very high change rate. 1PB of critical data over a weekend with zero customer issues. Not too shabby.

XCP delete is still our goto tool for massive tree deletes, and we do thousands of these per week with several dozen of these per week easily exceed 10M files. EDA, where file systems go to die 😦
A bit atypical but still interesting to note:
148M scanned, 126M removes, 21.3M rmdirs.
Total time: 5h42m

More typical: 20M removes, 906K rmdirs, in 47m.

Try with with rm -Rf and see how long it takes...

[hearty flax]

Hello everyone,

Can you tell me if the copy offload functionality in VMware environment works when copying data between two virtual machines?

[uneven roost]

It does not - only when it is between datatstores by the hypervisor

[hearty flax]

Thanks for the reply.

[hearty flax]

HI @uneven roost Could you tell me if using RAW device disk would work, in case Windows VMs make use of ODX.

[cedar raptor]

if you do a storage vmontion between datastores. an RDM does not need to move.

also, check out this kb - it might be helpful.

https://kb.netapp.com/onprem/ontap/dm/VAAI/What_VAAI_offloaded_operations_are_supported_by_Data_ONTAP%3F

but lets take this convo to #1063548056945037352

[hearty flax]

It would be copying data between virtual machines, I will look at the doc you sent. Thanks

[cedar raptor]

Got ya. The copy offload isn't aware of anything within a guest OS.

[pale cradle]

Svmdr between datacenters for cifs svm hosting user profiles - what do you think about it? On paper it looks great. Profiles in vhd containers

[crystal solar]

i'd think that vhd image consistency could/would be a problem... metrocluster or sync snapmirror would be necessary, but metrocluster would be best probably...

from what I know about fslogix containers, you're looking at considerably more IO than with profiles in simple cifs shares

[pale cradle]

There will be only 100 profiles. So io would not be a problem

[pale cradle]

Snapmirror in svm dr is not enough for consistency?

[crystal solar]

since a vhd file has it's own internal filesystem, essentially a lun file, and it stays open, I would imagine there needs to be a consistent state ... much like a hyper-v vhd file or a vmware vmfs file ... so simple snapshoting and mirroring may not get you there... trying to fire up your desktops from inconsistent files on the dr side could be problematic...

but i'm at the "theoretical stage" ... i haven't needed to do this... metrocluster takes care of such issues

i could be totally off on this not having done it myself or read enough... i'm not a big fan of such ideas generally... i don't see significant advantages in using disk blobs for vd profiles ...

[pale cradle]

its a little bit different - its citrix profile container in vhd file and its on logon copied to vdi and on logoff copied back depending on policies. it stores only user profiles and creates changes in different read/write files and on logoff from vdi it merges it into source VHD containing profile

[crystal solar]

still not going to sync with snapshot times unless your running syncronous snapmirror...

[pale cradle]

thanks for your feedback! will test it out

[hollow elbow]

Hi Just bought a NetApp fas2240-2, Can i factory reset it without needing to Enter a license key?

[cedar raptor]

nope, if you wipe it they're gone.

[vocal shoal]

Hi everyone, I’m bit new to storage (NetApp). Wanted to know if it’s possible to find out (via cli)which volumes reside on SSD vs SAS va SATA in Ontap 9.12.1P2 v

[vivid gull]

'disk show' will show the type of disks and aggregate names. 'vol show' will tell you which aggr each volume is on.

[vocal shoal]

Thanks John. Indeed, vol show works in my case. I was trying to get an output of volumes on each disk types, vol show does the magic. Can sort it out manually from there. Thanks very much for your help

[crystal solar]

you can do it via the cli also... just 'vol show -aggregate your_ssd_aggregate' for example

[vocal shoal]

That worked as well. I also ran a powershell cmd “get-ncvol -vserver mail* -aggregate n06_agg1_ssd, other ssd aggr” which gave me same results

[crystal solar]

if you mark your aggregates with ssd, then a wildcard will work as well, -aggregate *ssd, for example

[vocal shoal]

Yes, that’s exactly how I found out what SSD aggregates there are

[tired radish]

downloads are not available from mysupport for me and collegues, is there a known site maintenance in progress?

[tired radish]

Looks like its up again

[warm tree]

Does anyone know if a DS4246 with swapped out IOM modules to IOM12 will actually link up the disks at 12G ? I have this setup in my test environment, and I know that it works on ONTAP however it is hard to see at which speed the disks link up at... But in my home lab with Linux and 12G HBA, IOM12 in a DS4246 and 12G SAS drives, I can see that the drives negotiates 6G and not 12G... So maybe the DS4246 backplane can only handle 6G ? Can anyone verify this? If this is the case I would have to invest in two DS212C shelfs or the big boy DS460C...

[umbral stratus]

@warm tree that is correct. That was a SAS2 shelf, and to get SAS3 and the 12Gbps data rate you'll need a DS212C or DS460C for the 3.5" drives. The DS224C would be the 3rd option, but that is only for the smaller 2.5" 10K & SAS SSD drives.

You may need to also verify that the drives being used support 12Gbps rates.

[warm tree]

funny how NetApp has kept this a secret? 😉 As it is not stated anywhere? So basically the benefit of IOM12 would be the bandwidth between the shelfs to the controllers, but not all the way to the disks...

[umbral stratus]

I'll see if I can go look through some docs, but I don't think we intended to keep it a secret by any means.

[warm tree]

in my test setup with the Linux host in front it may actually make a difference with 24 drives installed... as the 6G backplane will become the bottheneck...

[warm tree]

don't waste your time on this. don't think I have any customers with that setup any longer... but it would have been nice to know as the time 🙂

[patent flame]

The DS4246 has a 6Gbps backplane, so even though it’ll run the 12Gbps modules, there’s still that choke point in the shelf. A little pointless to push the modules up to 12Gb unless you just wanna keep the connectivity and cabling current.

[vocal shoal]

Ports showing down on of the the node. Couldn’t find a command to check the transceiver power level for the port e0e but wondering if the command exist in the NetApp. Tried disabling and enabling the port, but the port is still showing down in the Netapp system manager

[dark sphinx]

Unless you have SSDs pushing GB/s on the disk shelves you should be ok with 6 Gb SAS.

Most FlashPool systems don't usually push that much to disks.

[uneven roost]

Yes - https://kb.netapp.com/onprem/ontap/hardware/How_to_view_SFP_optical_power_for_Ethernet_connections

See also https://kb.netapp.com/onprem/ontap/hardware/SFP_troubleshooting_workflow

[vocal shoal]

Thank you @uneven roost

[warm tree]

you are right, but if you have a mixed set of shelfs i.e. new DS212C and older DS4246 i would make sense because if you just use older QSFP cables it will drag everything down to 6G. This of cause only happens if you are trying to use older hardware, or did a headswap to a newer controller, keeping the older disks... and of cause NetApp would rather you just replace everything with new gear... so no evergreen here 😉

[drowsy dust]

Before I ask this, I understand the big performance delta with the disks, but otherwise.

C series to A series, is there a gigantic difference between them?
C series is newer hardware and beefier processors, but is that disk delta enough to reflect the proposed cost difference?

I'm comparing the A400 to the C800 and I'm not seeing much fo a drawback (for workloads that afford the shift from high perf to QLC)

[patent flame]

For me personally, I've found it to really come down to I/O density and sub-ms latency (from a disk iops perspective). Databases, large VMware datastores with intense production workloads, and funny enough, DevOps with constant build/teardown... those are the cases where I see people really benefit from A over C. TLC vs QLC def makes a difference, but it takes workloads that can truly beat the hell out of a box to illuminate it.

[drowsy dust]

Gen pop NAS where there are 8700's for NL-SAS based workloads (streaming cameras, big stupid pacs waiting uplift, imaging, etc.) That benefits from efficiencies like that.

Fabric pool workloads still perform about as expected with default offload policies?

This also, would be for a DR array, if it goes well we may consider it to replace an orange based piece of poop in a blade based orientation for immutable/indelible backup needs.

[patent flame]

That's an "ONTAP is far superior for that" discussion, not really a hardware one.

I wouldn't prescribe an AFF for that workload, personally.

[drowsy dust]

Agreed.

Thanks Nick:)

[minor acorn]

it takes workloads that can truly beat the hell out of a box to illuminate it
that would be us in EDA, although we also have workloads where the C filers would likely be adequate.

[dark sphinx]

C-Series only adds a couple ms of latency. Honestly most workloads wouldn't even notice compared to AFF.

[minor acorn]

For DR, I think C + StorageGRID would be a great combo

[dark sphinx]

If you have the budget for C series I'd say get it (over FAS).

[drowsy dust]

C + SGrid + FAS is looking like a winning combo for us so far vs the same but A instead of C

[cedar raptor]

C and A have the same hardware. e.g. A800 = C800 but with QLC disks.

[drowsy dust]

That's what I suspected.

Thanks for the confirmation Mike.

[unreal ingot]

Hi everybody, we are planning purchase of new NetApps to our on-prem DC mainly for VMware VMs storage. Previously we had planned to purcahse 3 x AFF A-250 (1 box with 24x1.92TB, 2 boxes with 12x1.92TB)... but recently our supplier told us there is new AFF C series. VMware is connected over NFS 3 and monthly average IOPS are 20K with peaks up to 35K (taken from ONTAP dashboard), latency average 0.7ms. We do not need low-latency access to VMs, we just need SSD-like access as we host a lot of developers VMs which are constantly reinstalled with new builds of software. Our old AFF-8040 which we are going to replace hosts ~100 VMs and 8TB physical used space (14.6TB logical). How can be AFF C-250 compared with performance to AFF A-250 and old AFF-8040, and do you think it could handle such performance as described above? we also looking to less rack space and less power needed and of course costs. Also does C250 have 25gbps ports possibility to autonegotiate down to 10gbps?

[cedar raptor]

Probably, work with your account team to run the older systems through fusion (our internal sizing tool).

[unreal ingot]

I can do that, but what about C250 performance? I saw intruduction videos, netapp onair videos, but none of them said, what is the max iops performance and what are latencies of C250 in comparison to A series.

[cedar raptor]

Docs say there's an additional 2-4 ms latency due to TSSE -

[sand flint]

Anyone using krb5p NFSv4 with Active Directory? AD Kerberos seems to have a maximum Ticket Granting Ticket lifetime of 7 days, and I'm wondering what folks are doing for users logged in longer than 7 days.

[dark sphinx]

C250 is fairly close in performance terms.

Should be 1-4 ms max disk latency.

[tulip raven]

hi Storage, I have a 4246 shelf and I'm looking to firmware upgrade the IOMs and similar. where do I begin?

[uneven roost]

Are you using it with ontap? If not, don’t bother

[tulip raven]

got it. thank you

[stoic jolt]

is there any update on ONTAP supporting STARTTLS to send email.. We are seeing a lot more customers not running on-prem email/relay servers and just using o365.. This KB does not say much. I know AIQUM support it.

https://kb.netapp.com/?title=onprem%2Fontap%2Fos%2FDoes_AutoSupport_support_Office365_as_a_smtp_server_%253F

[eager hatch]

May I ask why HTTPS is out of the question?

[stable grove]

I don't think this request is about how the AutoSupports arrive at NetApp. HTTPS is fine for that and I think most customers already use that. I guess this is about how a customer gets alarmed about new AutoSupports from a system.
Of course you can use other centralized monitoring/alarming tools like AIQUM but usually this still happens via mails directly sent from the system.
And as mentioned in the KB: "Microsoft Office365 typically requires STARTTLS for external e-mail destinations and it is recommended even for internal e-mail destinations."

[mossy ermine]

you can't send mails over HTTPS...

STARTTLS is an RFE that's still in development (see BURT 1098213). I heard rumors that it might be in 9.14 but don't hold your breath, the RFE is open for 3 years now 😉

if you have a support contract, I would suggest opening a ticket with NetApp and requesting to be put on the RFE. The more customers are linked to a BURT, the higher the priority (or so it is said)

[crystal solar]

'man kinit'

[dark sphinx]

The other option is the account team can push the ONTAP PM.

[dark sphinx]

Not really.

[placid creek]

AFF NVMeoF quick question - what does the 'delete-on-unmap' do when creating a nvme subsystem - I was told NetApp doesn't support deallocate/unmap (aka ssd trim) - or does this just mean if the 'mapped' host nqn is removed going to delete the subsystem ?

[tawny basalt]

Any chance you looked at the man page for that parameter?

[placid creek]

searched the wrong way - thanks for reminding me to search the root command - 'Specifies whether the subsystem should be deleted automatically when it is no longer mapped to a namespace.'

[plain kayak]

Hi everyone, I have two netapp aff a800, I want to replication. Can you help me?

[elfin igloo]

Hi @plain kayak You might want to consider submitting your question in the #1062049169520476220 Support/Questions section.

Also don't forget we have documentation @ https://docs.netapp.com

[cedar tendon]

[CVO/BlueXP on AWS] Can someone please answer me a quick question, or point me to the right documentation (I’ve been digging & can’t find a clear answer):

• Are CVO EBS volumes kept hot at ALL TIMES? Or are they meant to be snapshotted once they’re synced from on-premises? (This tells me maybe they can be snapshotted/pushed to S3, but not sure if there is something specific to be enabled here: https://docs.netapp.com/us-en/cloud-volumes-ontap-relnotes/reference-configs-aws.html#storage-support-by-license)
• Are there any other ‘on by default’ features that increase costs here? Or ‘off by default’ features that might help optimize?

[solid chasm]

In regard to XCP does it copy over file attributes and ACLs? Does it require to be installed on a separate host? Or if customer is migrating from traditional Windows File servers to CVO can it be installed on the File Servers?

[tawny basalt]

Two variants of the binary. One runs on Linux and copies Unix bits. The other runs on windows and can copy acl information

So yes, it needs a third party host. The more cpu/network it has the better. Years ago, I crashed a silly Solaris ZFS box using 7 threads over a 10g network

[solid cedar]

https://www.reddit.com/r/storage/comments/16ddpsz/netapp_aff_vs_pure_x90/

Any of you want to go drop some NetApp love in this thread?

[patent flame]

https://tenor.com/view/bait-thats-bait-tom-hardy-mad-max-gif-5055384

[gritty wren]

I'm pretty disappointed in you guys. Just got an A400 and C400 and still old bezels...

[tawny basalt]

Yeah... Same here. With all the pictures or there I expected a totally new bezel on the c800!

[patent flame]

@royal cedar may be able to weigh-in here with the situation. He is "Jeff Bezel" after all.

[uneven roost]

Short answer is that the first couple of months will still be the old “davinci” bezels instead of the new “capstone” ones. This was shared with our partners in the internal product brief

[mossy ermine]

nice, two more for my list of codenames 😄

[uneven roost]

The one before that was Sputnik apparently (FASx1xx and FAS[2,3,6]2xx) systems

[carmine lintel]

Anyone know what happened to XCPtion from Marko Haim? Gitlabs used to have it but that is gone and I can't seem to find anything about it anywhere

[solid cedar]

@carmine lintel I'm not familiar with XCPtion, and also found the same dead-ends that it sounds you encountered. We've got https://xcp.netapp.com/ if that's what you're after.
Nick has a good video on it too: https://www.youtube.com/watch?v=32OEtVIdJFQ

[carmine lintel]

yea, that's where I heard about the tool I was asking about, just seems to have vanished.

[solid cedar]

Let me message Haim about this and see if we can get him in Discord with us 🙂

[dire flint]

@dire flint Hi @carmine lintel DM me for XCPtion details

[carmine lintel]

sticking to the XCP questions.. is there no way to get more than one SMB agent going? we have multiple domains and I don't see them allowing me to spin up machines specifically for xcp in each domain.
hoping that xcption gets around that maybe.. no idea though.

[patent flame]

It’s really about the source and destination for XCP. Not necessarily the domains. A UNC path to another UNC path. It will parallelize the move. In task manager you’ll see it running 20+ “sessions” as it moves the dataset, depending on the resources available on the client machine. This is why it’s great to do it on a multi-CPU jump host, and the resources of the source and destination matter just as much. You can have multiple moves going on at the same time as well to consolidate to a single destination (ie Moving from a few diff sources to a single destination).

One of our partners, Komprise, also specialize in these kinds of large-scale data migrations if you’re looking for something a little more sophisticated with an elegant UI.

[carmine lintel]

I'm referring to the analytics specifically. It's the main purpose of us getting xcp going, we're trying to get file details for some shares that have hundreds of millions of files and break down their age and access times. but you can apparently only get one domain to work at a time

[patent flame]

Shoot me your email in a DM and I’ll reach out to the engineer for XCP and see if he can work on it with you

[stoic jolt]

We spun up a CVO instance for a customer yesterday and it automatically deployed 9.14.0. I cannot find any info on NetApp about 9.14. I thought 9.13 just got released.. we followed the bouncing ball

[mossy ermine]

9.14.0 is the cloud-only variant. 9.14.1 has not yet been released. I agree that communication about new cloud releases could be better, but it has been this way for a while now (9.1x.0 gets released about at the same time the first 9.1x.1 X-releases appear)

[stoic jolt]

I guess my question is if we hand this over to a customer and they go and look at NetApp support for any docs/info about 9.14 there is nothing..surely this should be a NetApp internal release before customers?

[mossy ermine]

yeah, feels like cloud customers are the new Betatesters these days. I asusme FSxN is also on 9.14 already? But I hope that at least NetApp support will know how to handle customers opening tickets for their 9.14.0 systems ...?

[stable grove]

yea it's a bit weird, no where any infos regarding changes in 9.14.0, you can't even download it, release-notes for CVO only mentions 9.13.1: https://docs.netapp.com/us-en/cloud-volumes-ontap-relnotes/reference-new.html

But it's surely not internal, it's mentioned in the ONTAP release-notes: https://library.netapp.com/ecm/ecm_download_file/ECMLP2492508
Also in the SnapMirror interoperability: https://docs.netapp.com/us-en/ontap/data-protection/compatible-ontap-versions-snapmirror-concept.html#

[gritty wren]

I did one final upgrade advisor today before upgrading one of my clusters (A300/A400) from 9.9.1P16 to 9.12.1P6, looks like 9.12.1P6 isn't available for download now and 9.12.1P7 is the recommended version. Anyone running 9.12.1P7?

[uneven roost]

You can type in a specific version usually

P6 was a specific release for one of our hyperscalar partners and is the same as P5 for all other purposes

[stoic jolt]

P6 only had 1 or 2 bug fixes.. P7 has quite a big list of fixes (some which we have hit).. I would be leaning to P7

[gritty wren]

That’s what I was thinking.

[carmine lintel]

in regards to XCP, who handles the security and vulnerabilities that are present for it?
we have some reported and our security team is pushing us to get some resolution and I can't find anything that indicates you can enabled/disable certain things in XCP.
XCP security vulnerability list:
Port:5030
Accepted:
Weak ciphers: AES128-SHA,AES256-SHA,ECDHE-RSA-AES128-SHA,ECDHE-RSA-AES256-SHA
Insecure ciphers: AES128-GCM-SHA256,AES128-SHA256,AES256-GCM-SHA384,AES256-SHA256,ECDHE-RSA-AES128-SHA256,ECDHE-RSA-AES256-SHA384
TLS 1.0 Weak Encryption Protocol
TLS 1.1 Weak Encryption Protocol

is this supposed to be handled by the OS ssl settings or is it done directly within XCP and it's own library/etc

[stoic jolt]

nice to see docs.netapp.com got some love with a polished update !

[frail cliff]

Hi All,

I am looking for some advice, regarding NetApp Data Erasure...

would this be the right place to do so? 😄

[cedar raptor]

I would ask in #1062049169520476220

(assuming it's an ONTAP system)

[frail cliff]

hey Mike, thank you for the response.

This is actually to do with fairly generic erasure

still the same place? its more the disks stored within said systems

[cedar raptor]

sure. ask away;.

[eager hatch]

Still no Dark mode though? So sad..

[carmine lintel]

anyway to speed XCP up when copying millions of files via SMB? can't use NFS. seems like it's much slower than robocopy/etc. It's pushing about 1000 files every 10minutes

[mossy ermine]

strange, usually XCP is much faster than e.g. rsync, but I've only ever used it with NFS exports, not CIFS shares. I guess what you can do is do parallel robocopy jobs (probably even from different VMs) to different subdirectories, to parallelize the load...

[tawny basalt]

The more cores the xcp host has, the more parallel processes that will run. The faster the network on the xcp host, the faster it will go.

In other words if the host has 4 cores and a 1Gb network it’s will be very slow. If the host has 16 cores and a 25Gb it will/should be really fast

[carmine lintel]

it's a 64 core machine, 256gb memory and 10gb nic
i have jobs running from 2 machines, one is going a bit faster, but it's still pretty slow.
one has pushed 20k files and the other has pushed almost 65k.
jobs started at the same time, been running almost exactly 6h. There are over 17mil files in total to move.

[patent flame]

@merry roost

^XCP guy will be here in a bit 🙂

[merry roost]

@patent flame @carmine lintel The linux and windows XCP's are very different internally. I created the XCP for NFS many years ago and can answer almost anything about it. Unfortunately the Windows version for SMB (AKA CIFS) is not reported to have great performance. Which you have confirmed 🙂

The Windows XCP was developed by another team so I can only answer high-level questions about it.

[patent flame]

Pete, would this large of a move be a good candidate for your XPack project?

[merry roost]

It's too small 🙂

Just kidding. For sure, XPack will a lot faster on SMB than XCP.

[patent flame]

Once we're past INSIGHT, I wanted to do a followup to my XCP video to go over XPack. Will be in touch soon to map that out.

Also, more on Haim's XCPtion project

[merry roost]

Perfect - see you at Insight or talk to you after

[minor acorn]

We're running P7 on multiple clusters without issues. Nice to see the aggregate size bump (from 195TB to 204TB on one of our clusters) amongst the rest of the normal new features and fixes.

[gritty wren]

Upgrading a cluster now, here, we, gooo….

[carmine lintel]

xpack? wazzat?
and yea, sadly we can only SMB for this as the system we are moving only allows SMB/CIFS traffic and not NFS, otherwise I would have mounted and be done.
right now I have 8 different VMs running xcp commands on different folders to get things moved as fast as possible. One folder is ~13million files and about 1.2TB of disk while another one is onlye 140k files and 120k folders.. 40GB or so. yea, it's a huge mess, glad I don't have to clean anything up, just move it to the new system
now just need to figure out how to get rid of the security failures on XCP so my security team will quit bugging the crud out of me about it
thanks for the info

[mossy ermine]

XPack is probably just the engine behind CloudSync? I remember there were talks about releasing that (or a limited version thereof) as an offline copy/migration tool back when it was first announced. I might be totally wrong here though 🤷‍♂️

[merry roost]

Early development of CloudSync used XCP as its copy engine when we got it started. There was a blog post by a customer about it a while back, so I am not divulging anything secret here 🙂

XPack is a new project I started from scratch right around the beginning of the pandemic; it incorporates everything I learned from creating XCP, and new ideas + goals + features.

It's in development.

[dark sphinx]

Couldn't you use Linux and SMB with Multichannel?

It's a little work to get it working but once you do it should be fairly stable nowadays.

[carmine lintel]

multi-channel is enabled, but it's still pretty slow with SMB. millions of small files just take forever, it has with every tool we've tried to use

[uneven roost]

I hadn't seen this before - https://www.tomshardware.com/news/western-digital-and-kioxia-to-announce-merge-this-month-report

[solid cedar]

Not to be confused with Kyocera. That would make way less sense. 😆

[tawny basalt]

Those wonderful Kioxia NVMe SSDs. I had a super micro I was playing with. Supposedly went through q/a. I didn’t believe it. After every reboot. Out of the 6 SSDs, you never knew how many would be available. They would just randomly disappear on reboot. To opening a case with super micro AND kioxia to get it fixed. I suspect drive firmware and ultimately got it.

[patent flame]

https://www.tiktok.com/t/ZT8B2Cbto/

The next few years are going to be LOTS of fun!

[granite flower]

I have a question about the NetApp 8040 Array. Two actually.

Is the NetApp 8040 Array the same as the NetApp FAS8040? And where can I find hte specifications for it? I need weight and dimensions for shipping.

[cedar raptor]

The FAS8040 and AFF8040 are the same chassis / controller. That Info can be found at hwu.netapp.com

I’d also recommend checking out he #1062049107096633454 section

[granite flower]

Much obliged!

[granite flower]

I have a question about transferring hardware ownership from one entity to another: I know NetApp's official policy is that they cannot transfer licenses, that it just isnt done. But I am not sure what precisely that means. The equipment has been removed from the original entities racks and is sitting on a pallet, we have a datacenter getting ready to receive it, I will be in charge of the equipment and fixing it and maintenance and so forth.

What does transfering licenses mean for me and for my company?

[patent flame]

@twilit pendant can you weigh-in here on what Jack's best course would be?

[twilit pendant]

This is a tough one. The end of Support date for the FAS8040 has already passed so you cannot get a valid support contract on that system. That will complicate you getting valid license keys for it. Did the previous owners wipe the system? There might not be much that we can do to help you here.

[granite flower]

We believe that they did not.

Its not our (my) first rodeo with EOL equipment so I am not worried about getting a valid support contract. Can the previous owners provide us with their keys? If so what are the consequences there for my company?

[cedar raptor]

I've seen this question pop up often on reddit and the community. @uneven roost or I usually reply with something along the following -

There's zero support for hardware as well as software - no upgrades / patches etc. the EoL on the system just adds to that.

NetApp Support or account teams wouln't provide the keys, the former owner will have to. (all though they shouldn't)

Fine for probably Lab work or dev, but I wouldn't run production on it.

[uneven roost]

Yep, the big issue is that we won’t provide keys and the previous owners aren’t supposed to either, but.. unless it’s being done at scale, I can’t see suing someone over it being worth it. But the license is reasonably clear on it - https://www.netapp.com/media/14114-enduserlicenseagreementworldwide.pdf

As the recipient of the keys from a third party, you wouldn’t have executed a contract with NetApp, so consider external legal advice for how the EULA might impact you.

All we can do is tell you the official line - licenses aren’t transferable.

[granite flower]

Basically we can't use those licenses to come to NetApp for anything official like firmware, parts, support etc, is what I am gathering?

[cedar raptor]

Correct.

cause it's non-transferable. and that model is EOL.

but more the former.

[granite flower]

Good shit, I'm going to appraise my company of this risk, we are familiar with EOL stuff anyway so it's nothing new but it's good to make sure we know what we are getting involved in.

[cedar raptor]

the really bad part that folks get themselves into is if there is a controller failure and it needs to be replaced. you or even the former owner coulnd't get the keys for the replacement board.

[granite flower]

So we cannot pull the controller in the head for one 8040 and pop in another unless we had the keys for that controller?

[cedar raptor]

yes. Keys are tied to the serial #

[granite flower]

I am guessing, since at this point it's fairly obvious I am going to be involved in the second hand market to maintain this equipment, that I should get keys before concluding any purchase deal for parts.

Will NetApp give us access to official documentation and KBs?

[solid cedar]

As long as you've got a valid contract somewhere on your account, you'll get customer or partner access to articles. A lot of them simply require a valid login to the support site/kb and you'll have access as a guest.

[granite flower]

Sweet.

I definitely appreciate yall answering all my questions. I'll probably have to relay some of this to the C Suite just so they know what we are doing and can account for it but it doesn't really seem to me to be anything worrying over.

Then again it's not my money so what do I know. Lol.

[patent flame]

Using EOL gear is great for sandbox and lab/homelab environments, and we're working on something for that coming soon, but it is never a good idea to use EOL gear to host production workloads.

[granite flower]

Well my company mostly needs it for virtualization lab environments for software dev.

Look. When I first showed up I rode herd on six or seven odd Mac Pro Server 2012s.

They were stacked on shelves in a rack, one on top of the other.

Maintenance and service was a... pain in my neck.

So I am very excited to finally be getting all of the tools I could ever want and then some.

[granite flower]

I have another question: How hard is it to reset credentials on everything for these machines?

IPs, credentials, etc?

[patent flame]

Absolutely want to help, but it’s really easy to lose track of specifics in this chat channel. Would you mind posting the specific questions in the #1062049169520476220 support forum (for software) or the #1062049107096633454 channel for hardware? That way we can keep track of all the moving parts (plus more eng’s and support folks monitor those)

I think I also saw a thread in the NetApp subreddit with similar questions. So it’s kinda hard to keep track of what’s being answered where.

[granite flower]

Roger wilco! 🙂

[uneven roost]

Yeah, the people who answer stuff on reddit are .. almost the same people as on here 🤣

[cedar raptor]

https://tenor.com/view/spider-man-we-one-gif-18212100

[uneven roost]

Mike and I live literally on opposite sides of the planet, so we do a short changeover in coverage about now 🤣

[solid cedar]

Well, Mike is literally sitting on my couch right now, so it might be a minute. haha

[drowsy dust]

No confidence! 🙂

[drowsy dust]

Oh it's the 8040 guy, any luck?

[granite flower]

Everything is going swimmingly. We have licenses, machines were wiped to factory default before shipping, equipment is gonna go in a Big Boy datacenter...

sniff

Im so happy.

[uneven roost]

I remember installing my first 8040. I almost dropped it on my foot. Watch out for the rails 😑

[granite flower]

They have a server lift and for liability reasons, insist that their techs are the ones who operate it.

So it wont even be my foot.

[drowsy dust]

I skipped from 3270's to 8200's, big leap

But yeah, they're hefty boxes, don't let people forget that they're heavy lol

[cedar raptor]

My 2nd (or 3rd?) F9K install went similarly.

Rear part of one of the rails popped out while we were sliding it in.

(I am glad we field stripped it before lifting it up!)

[granite flower]

I can feel my heart stop all the way from here.

[dark sphinx]

I see how it is @cedar raptor , you're too cheap to get a hotel even though it's paid for. 😄

[sand flint]

anyone familiar with this netapp disk shelf https://www.ebay.com/itm/204297997841?

im looking at these in the near future for an expansion to my home server

had a few questions if anyone can help me

[stable grove]

Never liked the DS4486. I would recommend to ask your question here: https://www.reddit.com/r/homelab/
Or maybe just search DS4486 there, you will find many posts.

[sand flint]

i just wanna know if i get a lsi 9500 16e card would it work natively with windows server 2022

or is there any special netapp stuff i need to do

[patent flame]

It’ll work great, but there are some caveats to the 4486. Double disk trays etc. I would stick with the DS4246. Your experience will be much more pleasurable in a homelab.

[sand flint]

whats the issue with the Double disk trays? i plan on loading 2 22tb https://www.newegg.com/wd-ultrastar-dc-hc560-0f48155-22tb/p/N82E16822234538?Item=N82E16822234538 at a time

i can show my curent server config as well if anyone is curious

[drowsy dust]

Those 2 disk 1 caddy got weird

[drowsy dust]

You prepared to "fail" or take offline two disks at a shot each time one needs service?
There's no sliding the caddy out and only doing one at a go

[cedar raptor]

IIRC- That’s it really. The os needs to be aware that there’s a second disk.

[uneven roost]

I’d encourage you to read up on other people using DS4486s in homelab situations- it’s problematic compared to the DS4246

They’re.. special

[sand flint]

Ok thank you

[granite flower]

Its funny how stuff can get quirky like that.

You would think hardware wouldnt have so many peculiarities like that but they do.

[drowsy dust]

Just wait til homelabs try and gobble up a DS460c

[patent flame]

The chia farms are all over them

[drowsy dust]

Pennies on the Gb? 🙂

Cha-cha-cha-cha-Chia!

[mystic sonnet]

Can a shelf have an id of "00" or it has to start at "01"?

[tawny basalt]

Absolutely. It just needs to be unique in an ha pair

[mystic sonnet]

cool thanks

[uneven roost]

Internal shelves are usually 00 and on some platforms can’t be changed

You can change the shelf IDs with ontap offline with no issues

[mystic sonnet]

Can I change the shelf/shelves id from the AUTOBOOT prompt or ontap (maintenance mode or not)?

[uneven roost]

Nope, it’s like setting an oven clock 🤣 You need to power cycle the shelves after changing it too

[mossy ermine]

there's a CLI command to change the shelf ID as well, but it's very well hidden and you still need to power-cycle the shelf. Can be helpful though if you don't want to teach the tech guy in the datacenter how to set the ID with the buttons: just set it remotely and tell him to power cycle the shelf with the flashing ID 😄

[granite flower]

https://tenor.com/view/ninjala-jane-hacker-hacking-computer-gif-20337624

Me setting my microwave and oven clocks after a power outage.

[cedar raptor]

outside of the annoyance of setting up. the stove i got last year has wifi connection and sync with a time server. it's kinda neat.

[minor acorn]

but does it do automatic daylight saving time changes based on your gps coordinates? 🙂

[mystic sonnet]

@mossy ermine I'm curious to know the command

[mossy ermine]

I hope it's okay to share here... the command is node run <node> sasadmin expander_cli <ha>.<shelf>[.<module>] 'set_encl_id <id>', for example node run cl1n1 sasadmin expander_cli 0a.10 'set_encl_id 23' to change shelf ID 10 on adapter 0a to ID 23.
The single quotes around the command are important, if you do an interactive node run you can also use double quotes (or maybe even leave them out completely). Also you can cancel the ID change by using the previous ID in the command.

there's also a help command for the SAS expander that lists some other (mostly safe) commands, but there are also lots of undocumented ones that can reboot your IOM, trigger shelf faults or corrupt the CPLD on the backplane, so test on non-production systems first 😉

[cedar raptor]

Should be OK to mention here. that command is in at least 2 KBs that are customer/public facing.

but those are related to when ONTAP thinks there's a different shelf ID vs what's been physically set.

i'd personally stick with the tried and true Alarm Clock style shelf ID setting process.

[mystic sonnet]

Hey it's good to know there's a cli-way to do it, thanks for sharing, doubt I'll need it much, but it's good to put in my notes; will experiment for sure; thanks again for sharing.

[patent flame]

Running away with it. Love to see it.

[carmine lintel]

couple of question regarding XCP.
I have an ask by our management team to pull a list of all file share data for customers. file count and size. This is all contained in one folder named "customerhome" and each customer has their own folder under that.
Is there any way to get that information without having to run an xcp scan on each sub folder?

second, I have XCP installed on a linux machine and running it on a windows machine for the analytics stuff, but when I try to actually look at the analytics tab it just blanks out and then returns to the home page.
Is there anywhere to get help for troubleshooting that? ticket to netapp or something?

[patent flame]

@merry roost

[uneven roost]

Have you looked at ONTAP file system analytics (FSA)? That might solve the first one

[dark sphinx]

Yeah you can open a case for XCP.

[stable grove]

No 9.14.1 release candidate today? ☹️

[mossy ermine]

that would almost be a first, releasing on the intended date 🙂 check again tomorrow, sometimes they're just 1 day off

[carmine lintel]

Yes, it does not provide information that is easily accessible for what we need

[carmine lintel]

I have no option to open a case for XCP, migration tools or such. I have asked one of my support contacts to see about opening a case for me

[long wadi]

Hello, I am having problems with a Fas 2040 controller, I can replace it with a Fas 2050, would it be compatible to have the 2 together in a NetAPP Fas 2040?

[dark sphinx]

Ah sorry. E-mail me if you don't make headway. My e-mail should be obvious.

[cedar raptor]

Hey Jorge, no for serveral reasons.

The biggest being the different form factor chassis

[mossy ermine]

FAS2050 won't physically fit into the FAS2040 chassis

[long wadi]

Thank

👍

[uneven roost]

You should be able to move the disks over if you’re running 7.3 on the 2040, but from memory the 2050 tops out at 7.3, while the 2040 can go to 8.1

2020/2040/2050 internal disks are cross compatible and both 2040/2050 support SAS shelves

I might be wrong, it’s not 2008 anymore though, so please forgive me 😛

[uneven roost]

this is pretty cool - https://www.reddit.com/r/DataHoarder/comments/17ypqin/laptopbased_das_netapp_ds_2246_on_a_laptop/ 🙂

insane, but cool

[cedar raptor]

wonder what the battery life is like. 🤔

[mystic sonnet]

I'm booted into maintenance mode on a 9.5 ontap on a fas2552 and I need to "resize" blocks from 512 to 520 and install a new ontap on this thing, do you know if it's possible to resize/reformat disks in 520 bytes per sector from ontap or I have to do it from a SAS attached linux box?
I suppose "scsi format" would work here?
cc @mossy ermine

[patent flame]

Well, two things, as I've been down this road before.

1. Yes, you'll need to attach the shelf of disks to a linux host and use the Seagate "sgutils" to reformat.
2. There's no guarantee doing that will work anyway. ONTAP requires the disks be on our Disk Qualification Package (DQP) which you must have updated in order to run it on the controller.

It depends on what you're trying to do, ultimately. Do you really, really want to run ONTAP? If so, I'd recommend finding some proper NetApp drives. If not, and you're ok with a consumer storageOS like unRAID or TrueNAS, then just leave them as 512 and attach the shelf to a separate host machine.

[mystic sonnet]

sorry, I forgot to mention they were once netapp drives.

[patent flame]

Ah! Perfect. Then they should work just fine. I'd be curious to see what ONTAP says about them in their current state.

[mystic sonnet]

0a.01.0 SA:A NETAPP X423_TAL13900A10 NA00 Y3C0XXXXXXX ff 1758174768 512 N

[patent flame]

You "should" be able to reinit the system and if they're valid/recognized drives, it should handle reformatting for you.

[mystic sonnet]

it says something like that...

cool

[patent flame]

Looks good to me

[mystic sonnet]

perf

[patent flame]

There's a lot of "it depends" when it comes to drives. We do have a small layer of support for some 512's built-in. Looks like yours are detecting fine.

[mystic sonnet]

yeah it should be fine, nothing a scsi format can't solve I think.

[patent flame]

do a storage disk show and let's see what you've got?

[mystic sonnet]

ah sorry it will have to wait until scsi format is done.

[patent flame]

no worries

[mystic sonnet]

I suspect 900GB would take like 2-3 hours no more.

[patent flame]

"it depends" 🤣

[mystic sonnet]

haha word

[patent flame]

are you doing this format within ONTAP or outside using a 3rd party tool?

[mystic sonnet]

ontap

[patent flame]

Perfect

[mossy ermine]

pretty sure you need to do it from linux. sg_format --format --size=520 --count=-1 /dev/sg<x> ... if your disk is from an E-Series or other enterprise storage system, you might also need to add --pfu 1 --fmtpinfo 0 (or something like that) to get rid of the protection field. But that only works for real SAS drives, not SATA (aka NL-SAS)

[uneven roost]

X423 is a sff 900gb iirc

[mystic sonnet]

@mossy ermine you're right "scsi format" didn't make them 520b

So there's just no way to do that from ontap itself? I'm not in the datacenter so I'd have to drive to connect it to a linux box (and throw an IOM6 in there too)

[mossy ermine]

did you use the --size=520 parameter?

ah you mean the scsi format in ONTAP? yes that won't change blocksize

[mystic sonnet]

yeah I on;y have ontap now, I'd have to physically put an IOM6 in that baby

so that'll be for next week

that's why I was somewhat hoping it would be possible to do it on ontap 🙂

[mossy ermine]

I recently saved 24 E-Series disks (10TB) from the trash, put them in an old DS4246, reformatted them with 512bps (and removed the T10-PI), and then plugged the shelf into an old decomissioned SolidFire node with a SAS card that I reinstalled with FreeBSD... Now I have a ~190TB ZFS array that I don't really know what to fill it with 😉

[mystic sonnet]

lool

I need to swap the fans on a 2246 to make it "office friendly", probably go with noctuas

I know it's been done by others, is there any documentation on which fans to get etc?

I suppose we only need to replace the PSU fans right?

I can't imagine anything else in there is noisy

I know it was done on 4246/4243 but idk if someone successfully did it on 2246

[stoic jolt]

Is there any feedback about running Virus scanning with ONTAP these days. Is the best option still using Fpolicy and paying an extreme amount of $$ to AV companies for integration.. Or do you get the same results with running AV endpoints (laptops/servers) and say mounting all your volumes onto a decent sever and doing a full scan weekly for AV? thoughts/opinions ?

[mossy ermine]

@mystic sonnet @uneven roost maybe we need a "Homelab" channel for such discussions ?

[solid cedar]

Use #┊・hobbies for now?
Nick and I have some ideas for a homelab channel already… 👀

[foggy carbon]

oh noooo! Will I have to spend even more money on homelab stuff?

[mossy ermine]

the trick is to get stuff that others are throwing out, for free (although I admit that doesn't really help against the continuous increase of homelab hardware piling up....)

[drowsy dust]

Which fpolicy vendor are you looking at?

[carmine lintel]

We have used McAfee, now Trellix, for years and it isn't that high of a cost. Works very well, easy to setup and configure, etc.
our reneweal cost is less than 10k/year

[lone slate]

Hello,
Does NetBackup supports Snapmirror Cloud? I couldn't find any information about that.

[mystic sonnet]

Could someone remind me how to get the size of a flexclone'd volume in 7-mode?

Like... I know the parent's size, I just want to know the size taken by the flexclone'd volume.

I did substract vol clone split estimate vol_name from df -m's output, but it's not even accurate.

seems to be rounded to the TB, I'd like to be a bit more granular

[mossy ermine]

can't you just look at the size of the base snapshot in the clone volume? That is, unless you have multiple snapshots there

[mystic sonnet]

and substract it from df -m?

hmm, not sure how I can get the size of a single snapshot (as opposed to the aggregate value of all of them)

maybe there's an "advanced" command to see the size of a particular snapshot?

if so I don't know it.

[mossy ermine]

I would assume that it's simply the sum of all the snapshots together. Because those are the differences from the original (parent) volume

[uneven roost]

https://www.reddit.com/r/homelab/comments/190zyo7/got_a_netapp_rack_the_other_week_my_mom_has/ .. this is cute

[vocal shoal]

Hello folks, I have a question regarding high disk utilization. There are couple of SAS aggregates that has very high disk utilization, between 95% to 100%. What I would like to know Is if there's a way to run the report to see which volumes have high IOPS and disk throughput. Please provide your guidance. Thanks!

[patent flame]

Drop a post in #1062049169520476220 forum, our engineers monitor those. This is more for general storage discussion and theorycrafting.

[vocal shoal]

@patent flame thank you. Posted in ontap channel

[reef elm]

Hey @patent flame, I’m rishabh bansal. A Netapp Presales in a firm. I know my role is technical sales but I really wanna learn the administration and implementation too. Is there a way I can learn? Sabacloud administration courses are paid.

[fallen mulch]

Try the labondemand... Trial by fire is the best and download the NetApp simulator as well...

[mystic sonnet]

A bit of a silly question, if I DELETE 1TB of files on a nfs volume, will snapmirror have to transfer 1TB?

trying to figure out why the snapshot delta is so high recently

[signal halo]

absolutely not!

[mystic sonnet]

I thought so, how can I investigate further what may be the cause of so much delta between snapshots?

just go in the .snapshot and diff the file differences?

based on file sizes?

[dark sphinx]

I feel like there is a better way, but yeah that might work.

I think 1 TB will send a bunch of holes. It will look like it isn't transferring much, but if you look at wafltop you'll see a bunch of repl punch hole messages.

[signal halo]

if you are deleting millions of files, although sm doesn't have to transfer any actual data, it still need to update pointers on the other side to reflect what you deleted. so that might contribute to some traffic, but if you are deleting a single file of 1TB, you wouldn't see much traffic at all.

[dark sphinx]

That is correct. Also if there are a lot of deduplicated blocks that will cause some churn.

[mystic sonnet]

Paul Stejskal: I feel like there is a better way, but yeah that might work.
What do you suggest, the problem with crawling snapshots, is that you really have to scan all the files for metadata and that's a lot of requests to the nfs head, it would take a considerable time (and possibly load)
An alternative would be to instrument things on the (linux) clients, not quite sure if there's something in this space

but we could use bpf to instrument write syscalls and others comparable ways, but that's going to be a lot of work unless there's already an opensourced solution in that space or some nfs instrumentation we can add to the clients.

let me know what you can think of
cc @dark sphinx

[dark sphinx]

I have no idea. Maybe the account team might be able to help you whip up a solution. The only other thing I can think of is maybe making a clone of the snap and using that.

[mystic sonnet]

it was worth the brainstorm

[quick lynx]

Hello everyone,

I have a question regarding performance tests in our environment - we're on vSphere 7.0.3 and have 2x AFF-A400 in two datacenters in sync

is this something for #1062049107096633454 or #1062049169520476220 ?

or here?

[solid cedar]

Let's go for #1062049107096633454

[fallen halo]

LUN question - with space allocation turned on and the lun has been turned off and back on, which procs ONTAP to perform the space reclaim? It is based off a schedule or a job? If so, is there any way to invoke that manually?

We have a transient LUN that sees lots of writes and deletes, but ONTAP isn’t seeing the avail space as the host sees it. We can’t resize the LUN any larger so looking to see if this is a disruptive move to a new LUN provisioned correctly or if space allocation can save the day.

[mossy ermine]

space reclaim only happens when the host sends SCSI UNMAP commands (or if the host writes zeroes that then get coalesced by ONTAP)

so maybe your host doesn't properly recognize the LUN as supporting SCSI UNMAP

[fallen halo]

Thanks for the reply. Turns out the issue was the host was rebooted, but the lun wasn’t actually turned off and back on. Once that happened, the space eventually freed up

[patent flame]

https://tenor.com/view/well-there-it-is-jeff-goldblum-gif-14058551

[dark sphinx]

Did you try turning it off and on again?

[cedar raptor]

https://tenor.com/view/it-turning-off-gif-19018125

[fallen halo]

I failed as an admin. Forgot the number 1 rule of IT

[brittle hearth]

So I'm trying to come up with a data strategy for a new server closet that I shouldn't be in charge of. Priced out a 2820(HA) intended for k8s snapshots and general file storage. I have an old synology at a separate site and was considering employing it as a disaster backup, but I'm not seeing a clean way to do this in documentation outside of paying for bluexp. Am I missing something easy?

[carmine lintel]

what do you mean? To do anything as a DR with NetApp tools you would need another NetApp on the remote side.
the only way to use something else would be via a backup software solution/3rd party tooling (rsync/ndmp/etc)

[brittle hearth]

sorry, my understanding of the netapp ecosystem is pretty limited. I figured there should be some way to send snapshot copies to another host.

[carmine lintel]

snapshots are netapp to netapp unless you use another external/3rd party tool to do any form of backup/etc. unless something has changed recently, which I haven't seen or read aboutj

[brittle hearth]

gotcha. Thanks

[patent flame]

We partner with Commvault/Veeam/others to facilitate these kinds of situations, typically

But as Ty said, easiest way would be having a NetApp system on both ends. If the remote is nothing more than a backup target, often times we'll see people go with a "single head" solution where there's no HA, just to save on costs. It's not ideal or recommended, and there needs to be some appetite and willingness to accept it going down if there's no failover, but if that's amenable, then you can save a ton. I did it as a customer, FWIW, and that second controller can always be added later.

Think of it less as system-to-system, and more about ONTAP-to-ONTAP. The model numbers on each end are mostly irrelevant as long as they are similar versions of ONTAP.

These are also great opportunities to consolidate whatever everything else might be running on. So don't lose sight of that either. The beauty and biggest benefit of ONTAP is the shared storage nature of it. By the time I was done (before joining NetApp in 2011) we had EVERYTHING on our NetApp stack. VMs, databases, home drives, departmental shares. Everything.

[stoic jolt]

Whats the current status of BlueXP getting multitenancy access? eg. Partners can be added/invited into a BlueXP tenacy ? I keep hearing reports its coming soon but i have not seen anything eventuate ?

[patent flame]

That’s an NDA thing you should ask your partner manager. 🙂

I’ll nudge em if you need me to

[fringe lava]

Any ideas on how to obtain our used unstructured data capacity across our landscape ? any guidance is appreciated....thanks

[cedar raptor]

just NetApp storage?

[fringe lava]

yes

AIQUM ?

[cedar raptor]

Yep. Aiqum is a good place to start.

[cedar raptor]

How many clusters?

[uneven roost]

Bit of a quick update - NetApp will soon offer 30.7TB SSDs for the C-Series systems. Available for quoting next week if anyone is interested!

[foggy carbon]

yeah! Let's see what the price tag says...

[mossy ermine]

ah right the embargo is over today 🙂

but yeah, that was the most interesting announcement from that presentation

[uneven roost]

It’s like Apple product announcements now 😉 it’s all so good, what else can we do? 😉

[mossy ermine]

I was hoping for some "just one more thinng.." style of announcement, and an image of a glass of wine or something 😉

[foggy carbon]

spoiler alert 😉

[crystal solar]

I'm probably just a crusty old fart, but I'm not a big fan of such "productifications" ... especially when it creates disadvantages for "bread-and-butter" FAS systems... it's cheaper to get high-capacity SSD disks by buying a C-series and tossing the controller in the bin than trying to upgrade from HDD to TLC/QLC drives on FAS systems

[foggy carbon]

well... we have a defined hardware lifecycle process of 5 years, so we buy the complete system... but yes, tossing thousands of ssds into trash that are well below write allocation is painful

[crystal solar]

ours isn't quite that strict, but i still can't quite get to the point where SSD makes sense for a "mid-tier" compromise... if you have fabricpool that takes the "dead" data, but having a "mid-tier" IO option would be nice. The gap between HDD and TLC/QLC drives is just too big still, last I looked. There are power savings to be had, but even that doesn't close the gap much.

[foggy carbon]

very true... the claims that disk is dead and ssds will be cheap enough for everything are at least 15+ years old...

[mossy ermine]

NetApp has a trade-in program where you can get discounts for returninng your used hardware. Presumably it then goes into demo systems, swing gear and/or NetApp-internal labs, which is all much better thann throwing them into a landfill

[foggy carbon]

yeah, but we have strict regulations... and netapp-internal labs are not a thing anymore (at least in Germany, where all offices were shut down)

[stable grove]

I know a "lab" which gladly takes your SSDs 😁

[foggy carbon]

Sorry, we have NRD🙈

[dusk compass]

hi everyone, I’m trying to add a hdd that already has files on it to windows pc. On the computer management I go to unallocated space and then new simple volume. Go through the steps and after selecting “do not format this volume” I get a message saying “the request is not supported.” any tips for this? Thanks. Don’t know the previous formatting to the drive.

don’t really do much hdd backups or tasks that involve storage. trying to help someone out with a backup and not sure if reformatting will cause any file loss?

[carmine lintel]

formatting will destroy the data on the drive.
Is this drive from a netapp system? Because if it is, there really isn't any way for you to connect it to a PC to get data off of it.

[dusk compass]

ok thank you, and no it was an external hdd used on a mac.

was formatted using hfs+ any tips I can do to get files over to an Exfat drive? Just got a usb to sata adapter will file explorer pick it up this way?

[uneven roost]

There’s a HFs+ reader for windows with a. Free trial - https://www.paragon-software.com/home/hfs-windows/

Or get sysrescuecd and make a bootable USB and move the files across under Linux, since that can read both exfat and hfs+

[dusk compass]

Awesome thank you so much!

[uneven roost]

Mostly for interest - newer systems use secure boot, but this smart person got unraid/linux running on a FAS3250 - https://forums.servethehome.com/index.php?threads/netapp-fas3250-hacking-unraid-installed.43638/

[patent flame]

I’m totally gonna try that with my older 2650

[mossy ermine]

I think you can even disable SecureBoot in the newer systems through the BIOS...

[rose thicket]

I think I am adding a new message

We are a closed shop with no access to cloud resources. However, we have developers who would like to test the waters with #object storage, S3 on ONTAP. Can anyone suggest where I can point them to learn how to use ONTAP and S3 object storage from a developers' viewpoint?

[carmine lintel]

https://labondemand.netapp.com/

[rose thicket]

Yes, I looked at the catalog, but except for a couple of courses for administration, aka StorageGrid and Astra Control Center, they're not what I'm looking for. I need to point my developer to how to use S3 object storage in their code, something which I am clueless about, of course.

[carmine lintel]

ah, the API/docs are good to start with
https://library.netapp.com/ecm/ecm_download_file/ECMLP2848267
https://docs.netapp.com/us-en/ontap/object-storage-management/index.html

also do a homelab via one of the courses from Udemy/etc
https://www.udemy.com/course/netapp-lab/
It walks through a full setup on your own system, allows 'most' of the functionality with limitations of course

https://mysupport.netapp.com/site/tools/tool-eula/simulate-ontap
netapp simulator

simulator does have a discussion/forum as well
https://community.netapp.com/t5/Simulator-Discussions/bd-p/simulator-discussions

[uneven roost]

https://twitter.com/davepl1968/status/1772042158046146792

Dave W Plummer (@davepl1968) on X

I wrote this Format dialog back on a rainy Thursday morning at Microsoft in late 1994, I think it was.

We were porting the bajillion lines of code from the Windows95 user interface over to NT, and Format was just one of those areas where WindowsNT was different enough from…

Twitter · Mar 24, 2024, 11:25 PM

[solid cedar]

He posted a video about that a few years back that I just remembered: https://www.youtube.com/watch?v=bikbJPI-7Kg

[sand flint]

Apologies if this is in the wrong section. With 7-mode we had the ability to simply copy the exports file so we'd have a backup of all exports. What are you folks doing with ONTAP since we don't have the same file?

[mossy ermine]

something along the lines of set -privilege diag -confirmations off -showseparator ";" -units raw -showallfields true; volume show; export-policy show and then import it into Excel

[sand flint]

Thanks for your response. Is this something you do or are you not concerned with it? Seems risky. I'm thinking some may leverage SVM snapmirror, which should include the exports, right? The good old days with the exports file made it easier to protect since you could make copies of that file

[mossy ermine]

what do you mean "seems risky"?

I do this all the time, autmated, on dozens of systems. I have some backend tooling to analyze the collected data for config issues and stuff (it's a permanent WIP though and it's not public 😉 )

[sand flint]

Risky in the sense to not do what you're doing or the fact that it seems more difficult to back up exports. I think you're on the right track and appreciate your insight. One admin's mistake via CLI could cause havok on exports without an easy way to restore exports easily and natively.

[frigid matrix]

Any other storage admins supporting an HPC environment? We're looking at a refresh, just looking for some perspective

Currently using an AFF-A800 backed by a Storage Grid for tiering off cooler data on-prem

[rose thicket]

@danko Yes, Windows HPC and LX Slurm, no tiering

[thin crest]

On cisco switches ,is it ok to do non-disruptive upgrade from 7.0.(3)I4(1) to 7.0.(3)I7(10) does it throw any surprises.. anyone got any experiences with non-disruptive method #┊・networking @tawny basalt

[tawny basalt]

I found this to be useful:

https://www.cisco.com/c/dam/en/us/td/docs/dcn/tools/nexus-9k3k-issu-matrix/index.html

I checked your scenario and there is an intermediate step. You will need to go to 7.0.(3)i4(x) where x is 7 or 8 or 9.

It’s also been a really long time since I’ve even tried the non disruptive method. Usually I have vPCs setup for everything and it doesn’t matter as long as one switch at a time is done

[thin crest]

Thanks for checking 😀

[tawny basalt]

If I recall it will check and let you know.

[thin crest]

Sure… 🙏🏼

[tawny basalt]

There were upstream checks and if it fails it says it will be disruptive

Just use that link to verify your upgrade path to mitigate extra surprises

[scenic pilot]

Looking for Support Engineer or Level 3 support with NetApp 9.3 Needed for troubleshooting work on 7 April Pacific time morning . Remote

[cedar raptor]

Have you opened a support ticket?

[rose thicket]

How much total storage are you managing on you HPC? I have about 6PB's and growing, and would love to compare notes

[hybrid ridge]

Need to provide disaster recovery protection for your VMware SDDC infrastructures? Are you using NetApp ONTAP storage? Then you should check out our latest BlueXP(tm) service called BlueXP disaster recovery.

Check out the latest entry in my ongoing BlueXP disaster recovery blog series to see how BlueXP DR work: https://community.netapp.com/t5/Tech-ONTAP-Blogs/BlueXP-disaster-recovery-How-it-protects-your-VMware-virtual-machines/ba-p/451695

[sterile quail]

hello, i'm Angelo, i need help for disk shelf de6600

[patent flame]

Hi Angelo, can you be more specific about what you need help with?

[pale cradle]

hey yall, I've seen this tool before - what is this? took screenshot but forgot who posted it. Maybe do you have some link for it?

[carmine lintel]

I'd be interested as well.
Xorux (Xormon/LPAR2RRD/STOR2RRD) doesn't look like that and it was what I first thought of..

[cedar raptor]

That’s fusion. Sizing tool for partners and NetApp staff

[pale cradle]

Thanks!

[stoic jolt]

Does anyone know when/if NetApp will start looking at Generic S3 eg. (Wasabi Object Storage) as another option for services like FabricPool, Cloud Backup etc..Quite a lot of our customers are now looking at it as it has a very compelling pricing model (compared to big 3) and there is only 1 tier. Also they have no cost to egress data out.

[carmine lintel]

good question, been waiting for that same option for quite a long time now

[crystal solar]

I can't imagine there are many upsides to such a plan. Licensing it in ONTap may well kill any pricing advantages because someone has to test, qualify, and support such an arrangement and these need to be supported by licensing fees plus some profit margin. These things are not small undertakings. It's only "cheaper" if you don't lose data or if your data isn't worth anything.

[carmine lintel]

It isn't always a matter of 'cheaper' but some people/companies have restrictions on where they can put files, and amazon isn't always on that list. Or, they prefer to keep all of their costs in one portfolio, etc.
Alternative options are always welcome, especially with the changing services being offered by all of the providers.

[dark sphinx]

No idea. You'd probably get a better answer from your account team.

[mossy ermine]

Anyone know what happened to 9.14.1P3? I cannot find any mention of it, even though P2 and P4 do exist and are downloadable. Was it pulled because of some issue? Was it never actually released?

[cedar raptor]

IIRC It was specific for a hypervisor.

[solid cedar]

^ That matches the notes in my email

[mossy ermine]

I thought 9.14.0 were the hypervisor images and the .1 were all for hardware?

[tawny basalt]

I heard it was a fix for aws

[cedar raptor]

.0 is for cloud only. .1 is hardware + CVS, CVO, FSxN

[carmine lintel]

.0 cloud, being what platforms?

[cedar raptor]

Cvo cvs etc.

[mossy ermine]

Ok now I'm confused. Is CVO/CVS .0 or .1 now? 😄

[cedar raptor]

Beginning with ONTAP 9.9.0:

ONTAP 9.x.0 releases are cloud-only releases and support Cloud Volumes ONTAP systems. The asterisk (*) after the release version indicates a cloud-only release.

ONTAP 9.x.1 releases are general releases and support both on-premises and Cloud Volumes ONTAP systems.

[uneven roost]

9.14.1P3 was released for a hyper-scalar partner who had a policy to not run D patches and who needed an issue specific to their environment fixed. For ALL other customers it is equivalent to 9.14.1P2

[lusty jackal]

9.14.1P3 was not publicly released. Public releases went P2 to P4.

[mossy ermine]

Yeah, I figured as much. It's just quite peculiar as I don't think this has ever happened with any other P-Release in the past (at least back to 8.x)....

[rose thicket]

A lot of announcements and porridge, but from what I see, not a lot of meat. Actually guys, I'm not interested in spending my time listening to people cheering the versatility of abstract or surrealistic features. When will we get some real meat and potatoes?

[mossy ermine]

what?

[dark sphinx]

You said CVO Mike which was confusing.

I believe he meant .0 releases are CVO only, but .1 is CVO+FAS/AFF (not counting others like OTS, FSxN, etc.)

[tall silo]

RFE for login scripts on netapp 😄 please!!! i'am so tired of row 0 and set d -c off in a big environment.

[tall silo]

BURT 500834, "RFE: CMODE. make set -rows, confirmations, etc a persistent option for users". Age: 13 years, 7 weeks, 1 day, 13 hours. 😄

[stable grove]

Doing "-c off" for every sessions is risky, I don't understand why people do this. You could f#ck up so easily especially if you use * in your cmds. I only deactivate the confirmations when necessary.

[tall silo]

and stuff like that. And tbh it's just an example but hell i'am here on at least 120 controllers and no full automation for some basic Ops work and with cyberarc i'am on 3 lines to get on a system and then 2 additonal lines before i start using commands gets just annoying 🙂

[mossy ermine]

agree. I also don't understand the affection towards rows 0 as it toally messes up column formatting which makes the output extremely hard to parse visually (e.g. snapmirror show) . I usually prefer rows 9999: same effect but keeps the formatting intact

[tall silo]

well that could be your choice in your login script 😄 when netapp would give it us. I prefer 0 🙂 except ofc for logfiles but most commands i want just rows 0. I dont like line breaks and you still have them with rows 9999 and it's even more to type 🙂

[mossy ermine]

I guess I just never unerstood why "rows 0" also completely breaks the column formatting and not just disables paging... this should have been two separate options

[atomic cargo]

Some users uses rows 9999 as a workaround

[mossy ermine]

yeah that was my point above 🙂

[atomic cargo]

Oh

[tawny basalt]

I use set diag, rows 0, specify fields and then one or more sort-by fields.
Great tabular format every time

Net int show -fields home-node, home-port, auto-revert, service-policy, address, netmask -sort-by vserver, lif

[mossy ermine]

...but that still messes up column formatting right?

yeah, it does. might be nice for consumption by a tool, but for visually spotting anything unusual, it's terrible (IMHO)

I mean... yeah... it's more terse but the verbose one is easier to parse, at least for me

[vivid gull]

rows 0 works for somethings where line breaks make it a pain to parse. I run into that a bunch when the node names are too long

[mossy ermine]

for parsing I usually do -showallfields true -showseparator ";" which basically gives you CSV (actually more like SSV, semicolon-separated values but it imports in excel or any tool that can parse CSV perfectly)

I actually have a small shell script which collects a complete diag log of almost everything in that CSV format, and a corresponding desktop application which parses these CSV logs and detects config-issues

[tawny basalt]

The right picture has perfect column formatting just too many columns to be useful

Drop the service-policy. Either way, I’m usually looking for something. Getting the columns I need and then sorting makes spotting something easy.

Iscsi initiator show -fields tpgroup,initiator-name,igroup -sort-by vserver, igroup, tpgroup

This shows all connected iscsi clients, sorts by vserver, then igroup, then tpgroup. I can verify multipathing easily for every igroup per svm and which interfaces they are using. Easy to find what’s missing.

[carmine lintel]

which app do you use for parsing, or is it a public tool that can be shared?

[mossy ermine]

either Excel (with the "text to columns" button) for some quick overview or analysis, or in a custom tool written in c# for more in-depth analysis

[civic kelp]

We have a C400, with one NS224 disk shelf that is full (24 x 14TB disks). We are currently at about 75% capacity. If we purchase another NS224 shelf, what are the options as far as adding disks to that shelf? (Minimum number of disks, number of disks to add in one increment, etc.).

[whole saffron]

Correct me if I'm wrong but I believe once you get past the initial config of 8 drives, you can add 2 disks (Drive packs are sold as 2 disks) at a time after that including a new shelf.

[lusty jackal]

dude, Im the exact opposite haha. I can't stand not using rows 0. Also, it makes the data so much easier to copy out to excel. But I also always use the -fields and know what I want displayed because a lot of the defaults is unnecessary information.

The key is for those that use specific set values, they should be persistent. If you like them, use them. if not, use what works for you. I agree, it's a pain to set every SSH session to what works for you every time.

[carmine lintel]

yes, settings should have some form of save capability. sucks to have to set things the way you want every single time you log in to each system (we have over 20 clusters, 40+ nodes). setting them each time is annoying

[hushed birch]

Hi Folk. Here to pick people’s brains and see what other options we have.

So we have abit of a conundrum, A900 using SAN. We have a 40TB parent volume that we use as a source for flexclones, we previously created 8 clones off this parent and then subsequent child clones from them. We had an issue a while back where an end user did something on the VM of the parent volume which in turn cascaded down to everyone and in the end resulted in us running out of space.

To avoid this scenario we decided it would be better to split the clones from the parent so the impact would not cascade down. The issue we’re faced with is the split takes around 4 weeks for one volume, which is just not acceptable for us we run this process weekly. We’ve tried every tool available to try and get this data copied in a timely manner on the same aggregate with little luck.

We tried Lun copy, the cutover process was a 4 day process, snapmirror is just abit clunky and results in not having the copies available until we can break it. Vol move is not an option as we don’t have the free space in another aggregate to then move it back again.

What ever happened to vol copy in 7mode, we just want a copy of the data quickly and space savings can be post process but we’re finding it neigh on impossible to achieve this when it should be fairly simple. Has anyone else got any ideas

[mossy ermine]

vol copy is no more, it has been buried and lost to time 😉
Did you actually let the vol clone split run through? Or did you start it, saw the estimated duration was 4 weeks, and then decided to abort it? Because from my experience, the estimates for the split vary wildly abd in practice it will usually speed up a bit.
But for 40TB of data you are still looking at over 1 day if the system serves other data concurrently (as this will be prioritized over clone splits), so I think this is still the best option (because it gives you only a small time window where changes to the origin volume can "explode" and take too much space in the aggregate)
However, in the end, as with any thin provisioning, you cannot escape the physics of the aggregate. if you have 20 thin provisioned volumes of 40tb each, and people start doing massive changes on them, space usage will blow up to the point where you might need 800tb to store it all.

[hushed birch]

it was left to run for a week and 5% complete, split is just not an option as you cannot create snapshots on it until it completes. We understand that space is finite but spliting them from the parent removes the existing risk we have, it is not a risk for the splits because the process that was run only ever runs on the parent

Its a real shame because we've been netapp users for over 15 years and well versed in its capabilities, but not giving me the option of a vol copy or similar is poor, the tools are there to do this but just not exposed to end users. Instead we're hamstrung having to wait weeks for something that could be copied in a day or so.

For reference the system is 10% utilised on an A900 and we can't get it to improve... its all abit rubbish

[thorny maple]

What do you use for SSH? Where I work, we use MobaXTerm. With it, we can do macros that run every time you log in, so I have "rows 0" run every time I log in without even thinking about it. I'm sure other terminal programs offer the same ability (though I'm not aware of any way to do it with standard Putty).

Btw, IMO, I prefer the "rows 0" over the "rows 9999". I prefer to have all the output on a single line over having it line up in columns. But, that's just me.

[carmine lintel]

we use putty, and yea rows 0 all the way.
Regardless of a program being capable of it, should be an option to set in the system directly

[mossy ermine]

I'm not sure I understand your usecase. Why split the clones and not leave the clones as-is? If you only use 10% of the system it's nearly impossible to run into an out-of-space condition unless you change all the data on all the clones within a day or so (before your monitoring picks up the increased aggregate usage), at which point I would rather "fix" the user who does that instead 😉

But in any case, there is no vol copy command anymore, the only means of doing what you want is either a snapmirror, vol move or a clone-split 🤷‍♂️

[tall silo]

i still prefer the left side 🙂 and yeah often i go with fields and then i prefer 100% row 0. It's a matter of taste how you like informations served and i can't deal with line breaks.

[mossy ermine]

fair enough. I usually stick with the defaults (not -fields) most of the time because it takes much longer to type (even with tab completion, especially since it's not always clear what the each field is called), so I'm usually faster just seeing everything neatly aligned and then visually scanning for the "one-off" entry that is incorrect (or whatever I'm looking for) because that sticks out 🙂

[hushed birch]

sometimes you can't fix the user, this is the issue and just explains this could happen again. We can't restrict his access and this is just a genuine human error, the number of envs hanging of the clones makes it just too risky to continue this way. I know vol copy went away i'm just asking why can't I just have a full fat copy of my data rather than needing to split a clone. Its a fairly simple request wouldn't you think?

[drowsy dust]

SecureCRT used to allow auto-login scripts ages ago, I'm sure it still does.

Row 0 unless looking at logs, then row 40.

[mossy ermine]

can anyone explain (or link me to a doc that explains) what the security us-eo-cybersecurity modify command is used for? Or what kind of "alerts" it enables?

[patent flame]

@leaden garnet

[carmine lintel]

is there a way for XCP to keep scanning when it runs into an error?
I've tried several ways to do a scan on a set of shares and because some of the files get deleted after the scan starts it errors out with the "no longer accessible" and dies.
This is after running for 2+ days each time.

[pale cradle]

Regading xcp - can it be destructive on source like robocopy? Or it will just leave source intact or are there parameters we should avoid?

[patent flame]

@merry roost should be able to help here. I've never seen it be destructive. I believe the order I used was /scan, then /copy, then /sync, then /scan again, then /sync again, rinse repeat until it was completely sync'ed over and #files and size matched.

[pale cradle]

its XCP on Windows host to migrate CIFS

[patent flame]

Yea that’s what I used too. Whats the source and destination?

[pale cradle]

source cifs on fas2650 destination cifs on c250

[cedar raptor]

Why not snapmirror?

(Sorry if I missed that answer previously)

[pale cradle]

Yep i'd do that but source is on 9.7platest and dest on 9.14.1platest

[cedar raptor]

can't upgrade the 2650 to 9.9.1?

[pale cradle]

Out of support 😦

Contract for it ended

[mossy ermine]

don't you have some ONTAP update file lying around anymore? I mean, it could "accidentally" find its way onto the system overnight, or something 🙂

[pale cradle]

Ofc we have access to all newest ontap software, we have support for 9 clusters atm

[mossy ermine]

the 2650 can go up to 9.11.1 even

[cedar raptor]

I have worked with customers in the past for exceptions like this to make migrations easier for older platforms.

all data is planning to move off the 2650 to the 250 correct?

and then 2650 decom?

[pale cradle]

yep

[mossy ermine]

in that case a temporary 4-node cluster would also be an option

[cedar raptor]

also an option. Talk to your account team about getting that 2650 to 9.9.1+ for an easy snapmirror.

[patent flame]

SM is def the easier option here

[pale cradle]

yeah but it might be overkill. cifs is like 5TB of files, c250 is switchless

[patent flame]

I thought we might be coming from 3rd party

[cedar raptor]

only * on that is if it was moving to FlexGroup , in that case. yeah. have to re-ingest with xcp.

[mossy ermine]

easier than vol move?

[cedar raptor]

that's if they have switches.

[patent flame]

Well… yea that

[cedar raptor]

which most customers don't have a space pair laying around

we kept a few pairs of 1610s around for things like this.

when i was PS for a partner

[mossy ermine]

true. but at the end of the day you can use just about any switch for the migration. especially if the system is out of support anyway 😉

[patent flame]

I’m not in sales or support. I would just update ONTAP and let it SnapMirror, cut over and decomm. Gonna be easier than watching a cmd prompt tick by on a winbox using XCP

[pale cradle]

we will try on monday XCP, whole cifs is like 5TB so I think its nothing special

[mossy ermine]

yeah, we also have a dozen or so CN1610 for this reason. the problem is these are by now also unsupported on newer models, so we're throwing them out one by one now ...

[patent flame]

50TB took me about 3 days over 10Gb when I last did a big move

[pale cradle]

it will be 5TB over 1Gb (yep.) snapmirror network is also 1Gb

[mossy ermine]

XCP copies client-side so the smaller the files are the slower it will be, especially on SATA

[pale cradle]

yeah I know

so we might end with snapmirror

if this will fail

will give xcp a shot tho

just wanted to check if it will not break source 😄 like robocopy can

[mossy ermine]

unless you mix up source and destination it won't (robocopy won't either 😉 )

[pale cradle]

haha yeah but robocopy does have move and can delete on source afaik

[mossy ermine]

um, yeah, but so can Explorer 🙂

[pale cradle]

haha exactly

[cedar raptor]

snapmirror just makes it easy. Don't have to worry about permissions getting weird or something.

[mossy ermine]

yeah, and no need for backup mode or inaccessible files due to wonky permissions or things like that

[pale cradle]

I wonder if I can try to connect them or just it will threw error that ONTAP is to high on destination

[cedar raptor]

I'd go the ask the account team about getting the cluster up to 9.9.1+ unless you have your change for this weekend or something.

[pale cradle]

nah nope we have time

[tawny basalt]

Ndmpcopy is another option. Version independence

[pale cradle]

By any chance you guys know if we've got support active for 2650 in 10.12.2023 we can upgrade to versions and patches released before that date?

[mossy ermine]

only if you already have the ONTAP versions downloaded (or have at least one other system under support that enables you to download ONTAP installation files)

[pale cradle]

FYI guys even if its listed not supported for snapmirror from 9.7 to 9.14.1 it worked like a charm. We've already snapmirrored 1 medium share and it retained permissions etc. Now we started some large vols and it maxed out intercluster traffic.

[uneven roost]

We are starting to talk about svm migrate for shared nothing migrations

[stoic jolt]

Anyone know when NetApp will standardise on Linux distributions for their software ? BlueXP Connector: Ubuntu 22.04 LTS,Red Hat Enterprise Linux , ONTAP Mediator: Red Hat Enterprise Linux,Rocky Linux, Keystone Collector: Debian 12.. and im sure there might be more 🙂

[mossy ermine]

I see where you're coming from, but in the end, it's an appliance. What Linux it runs on should not matter to the user/admin, because updates are supplied/managed by NetApp.
This is a bit similar to customers complaining that MetroCluster uses Cisco/Broadcom/nvidia switches in the backend. "Out network team is not happy with that since we use (HP/Arista/Huawei/...) switches exclusively". It's a black box that they don't have to touch/manage, so why should it matter

We had this discussion (about the Linux distros at the base of various appliances) a couple of times, but that argument of it being an appliance and nothing they would have to (or even be allowed to) manage/fiddle with has always worked

[stable grove]

Wellll.... you do have to manage and patch the underlying Linux distribution. Only the software-part is the black box / appliance which you shouldn't need to manage. The OS part is still your responsibility.

"Maintaining the operating system on the Connector host is your responsibility. For example, you should apply security updates to the operating system on the Connector host by following your company's standard procedures for operating system distribution." (copy-paste from BlueXP connector docs)

If every of these different software needs another distribution your admin crew needs the skills to manage and secure all of them. So I sort of agree. Why not standardize on RHEL and Rocky?

[mossy ermine]

Because it's an endless discussion that has no correct answer. "But we have a SUSE support contract, why RedHat and not SUSE?" ... "All our Linux systems are Debian, can't we use that?" ... it gets worse if you factor in other vendors' appliances: "But vendor XY has standardized everything on Ubuntu, can't you use Ubuntu for everything?" etc. so yeah, in the end you pretty much have to just deal with multiple distros. In your case RHEL might be the optimal solution but I can find a dozen other customers who prefer anything but RHEL (especially since RHEL requires a support contract for production use)
There is only one possibility: containers. But many customers have neither the infrastructure nor the experience to run those properly yet

[stoic jolt]

I would love all these to be a NetApp provided appliance.. just like Aiqum and the VMware tools., take the whole, patching, support of a Linux host out of our hands

[stoic jolt]

100%, RHEL and if you want free Rocky.. but I’m also down for NetApp to provide appliances for all these..

[mossy ermine]

Isn't Rocky in (legal) trouble because they're using RHEL-patches against the Red Hat EULA (i.e. using a RHEL account to download patches and then applying these patches to their own code)?

[carmine lintel]

In the end it is, but isn't, an appliance, but your security team doesn't care.
When things are not passing security audits (which we have ran into numerous times) it is up to us to fix those.
It greatly matters to the admin because they are the one that gets the notification that something failed to pass security and have to find a way to remediate.
It might also matter because some teams are prevented from using specific software, such as us.. we can't use CentOS.

Even an 'appliance' like NABox I have to go through and make changes to for our security audits to pass. Not having the same set of documents to fix problems across all the software packages is cumbersome and tedious.

have to fix specific things with XCP, ActiveIQ, NABox and BlueXP. Each different because of the platform it's installed and running on.

[dark sphinx]

What you could do is request Yann to fix whatever security problems there are with NAbox.

(like 4.0.3 fixes the OpenSSH vulnerability)

[carmine lintel]

I have, and he did for the version we are currently running
Not had the ability to upgrade to 4 yet

[mossy ermine]

A security mitigation can be as simple as adding firewall rules so that the appliance can only originate traffic, and just close ssh/http/whatever to everyone (or everyone except a few select management workstations/VMs). But yeah, I get the "it has always been done like that, it will be done like that, end of discussion" argument, I have heard that much too often myself 😕

Maybe I should feel lucky that our security team is actually actively reading the CVEs and assessing if they even apply to us, rather than blindly force updaets on everyone for every single CVE out there. I'm expecting a lot of good things to come from Linux's decision to label (almost) everything as a CVE in that regard 🙂

[dark sphinx]

That's good.

[stable grove]

ONTAP 9.15.1 is GA! 🎉

[mossy ermine]

Admit it, you have a script or something that checks the download page every few minutes 😉

[stable grove]

I have this, you don't get the mails? 🤓

[mossy ermine]

yeah, I am subscribed but I don't remember ever receiving any mails... strange. I'll try to reset it

[icy oriole]

Yeah, just add a management-ssh service on a lif with -allowed-servers configured to one IP (from which you could get to it, but nobody else can)

(net int service-policy add-service -vserver [name] -policy [name] -service management-ssh -allowed-addresses <ipaddress/32>)

[icy oriole]

Did a few migrations with ndmpcopy to portable drives/FedEx back in the mid aughts. 🙂

[waxen frost]

Just installed without issues like to RC2

[solid cedar]

https://youtu.be/syVVvPFxSlQ
Let me know if it’s not working as expected. I know the guy who made the video.

[mossy ermine]

I unchecked everything, saved, and re-checked the checkboxes, let's see if this fixed it 🤷‍♂️

[stoic jolt]

This is getting commical now.. for an On-Prem BlueXP connector i need Ubunutu 24.04, Docker between 23 and 25 and a Datasense Collector i need Ubuntu Version 22.04, Docker version does not care (at this stage but ill use 25).. This is sooooo painful... We need Appliances of these from NetApp !

[mossy ermine]

The BlueXP connector supports 22.04, not 24.04, at least according to the docs...

[stoic jolt]

acutally your right.. ive been on so many NetApp document pages trying to figure this out.. I managed to get it working on 24.04... at least the installer does not check unlike the data sense installer.. it bombs out with 24.04

[dark sphinx]

Blah you're fired. This was posted in the Blue XP channe.

Just because it works doesn't mean it will continue to work.

I hate to be that guy but I'd suggest to install on Ubuntu 22.04.

[pine elbow]

👋 Hey there, I'm looking for some documentation or at least some experience reports regarding NetApp + ProxMox, i saw that there was recently some activity on the proxmox forums and i really look forward to see if there is some steam picking up in that direction, as we are recently trying to pull up a PoC for using ProxMox with NetApp aswell - Thanks!

[mossy ermine]

here is a guide on how to configure NVMe/TCP with ProxMox (which is quite simple since ProxMox is basically Linux)

[pine elbow]

Yeah I think that was the post that was linked in the proxmox forum post, so I already read that one, but thanks for pointing it out 🙂

[patent flame]

Check #╰・virtualization I just posted an update in there. If you check the Reddit post I linked, there’s an update this morning

[pine elbow]

Sweet! Thanks for pointing me to it, I'll check it out right now 😉

Nice nice, so it seems we're actually going in the right direction with getting our PoC setup up with NFS, sweet 🙂 Some points that still hinder our operations team to fully embrace the proxmox solution is, that they really want some UI way like in vSphere to quickly restore backups, seems like this is still underway with the Storage API rewrite of ProxMox, so fingers crossed, thanks for keeping up posted on the progress ❤️

[patent flame]

Correct, and we're likely a ways off from that. 1) we have no idea how long it'll take PVE team (Max) to get that over the hump and 2) we'll have our own eng efforts to solve for beyond that.

What I'm most excited about is being able to build a SnapCenter extension once we get the interop qualifications done. TBD, will keep everyone posted!

[stuck compass]

i have little problem maybe some 1 can help me i'am trying to update frimware of p420i raid controller but which is 7.0 to 7.0 (E) and give me this error Update not needed. All selected devices are already up-to-date.

Hp Dl380p gen8 server I try to update from esxi host whit ssh

[cedar raptor]

I’m assuming you’re using a DS4246 or something close for a home lab? You might get better replies on Reddit.com/r/homelab.

[warm tree]

Not sure if this is the right place to ask this, but has anyone else heard rumours of pricing going up on C-models? Because I have an existing quote that I have to change a bit, and suddenly the price went way higher than before... My dist. guy claims that the prices have increased the last month or so... and I have no way to check if this is true or not...

[carmine lintel]

not sure if it's just NetApp, but prices across all forms of SSD/NVME/etc are going up due to several things in the market. Some are suggesting that prices will continue to climb for the remainder of the year

[warm tree]

well I have a valid quote that has some c800 and som c250 in it. my customer now wants to change and optimize their setup by taking away a bit of disks on one system while adding some on others. But when I ask for this quote change, the overall price is much higher than the original quote, even though it should be a little lower... it's hard to explain to a customer that prices change seemingly without warning... now I am just a reseller, so I don't get that much insight.

[tawny basalt]

I hear a paartner communique is forthcoming. We should be aware as soon as the rest of NetApp since we are the ones facing the customers!

[stable grove]

Some info is available already in fieldportal but the whole news are coming start of next week.

[warm tree]

I have no problem with the prices increasing. It's just hard to explain to the customer that the little change we want to make which should make the quote less expensive, actually made it more expensive... (no warning from NetApp or distributor)

[stable grove]

The pricing changes did not yet happen. They only go in effect next fiscal quarter.

But please check yourself in fieldportal, as always this info is not public

[uneven roost]

Speaking personally, without reading the communications - my guess would be that quote validity extending into that period may reflect the changes - I’d say the answer for your customer is to buy from the original quote if it’s still valid. Unfortunately the market for flash memory is trending upwards for the first time in a while.

[coral agate]

Hey guys! First time customer of NetApp, just purchased 2 x C250 AF’s to host a virtual environment of 18 ESXi hosts. Hoping setup is relatively straightforward 🙂

[cedar raptor]

there's lots of docs out there to help and always feel free to post in #1062049107096633454 or #1062049169520476220 with questions!

[patent flame]

Amazing, and welcome to ONTAP! You're gonna have a blast. Choosing it for VMware in 2007 was the best decision I ever made. The efficiencies (even then) remain unmatched in the industry, and once you realize you can do SO much more than just vmware datastores on it, you're gonna unlock something you've never been able to do before. I went through it, and all those new to NetApp go through it. By the time I was done, we had consolidated vmware, backups, home drives, dept shares, remote sites, oracle mounts, exchange mailbox stores.... literally everything was in ONTAP, and changed everything about how I centrally managed storage.

[coral agate]

Thanks guys! Should be good, I've heard about the storage efficiencies and that's one of the main reasons why we opted for NetApp over a traditional Dell SAN.

[waxen frost]

There is much more then only the efficiency 😊 the whole automation with Ansible is much smoother. The monitoring topic with Harvest and the communication over Discord and Git is gorgeous

[coral agate]

Defo opening up more opportunities! This is good to hear!

[solid cedar]

https://www.reddit.com/r/todayilearned/s/DD1B205if0
Anybody want to talk F1 data storage on Reddit?

[spice flax]

Any good Cluster hardening documents available?

[patent flame]

https://www.netapp.com/pdf.html?item=%2Fmedia%2F10674-tr4569.pdf

That is the last publication of the official TR. Any deltas since will be in our official docs, as we've (mostly) stopped publishing these monolithic PDFs.

https://docs.netapp.com/us-en/ontap/ontap-security-hardening/security-hardening-overview.html

[stable grove]

Followup soon™ ?

Noticed something 👀
https://mysupport.netapp.com/site/tools/tool-eula/npack-toolkit

[mossy ermine]

I already have spotted a few issues that I could report 😂 but it's not on GitHub yet sadly

[crystal solar]

why is RHEL 8.3+ listed at supporting nconnect when not even a modern RH 9.3 release has a kernel version (5.14) that supports nconnect? Only kernel versions >= 5.3 support (nfs) nconnect

[mossy ermine]

5.14 is newer than 5.3 ... And even if not, RedHat is known to backport many things into their kernel from newer versions

[crystal solar]

looks like they backported it to 4.18.0-240.el8

requires nfs-utils-2.3.3-35.el8 libnfsidmap-2.3.3-35.el8 or later also, it seems... (there's your problem TM)

[mossy ermine]

ONTAP 9.16.1RC1 just dropped btw 🙂

[icy oriole]

Cool. [gets back to planning upgrading our metroclusters to 9.13.1P12]

[stoic jolt]

any ETA when NetApp will update its Visio Stencils for Nov hardware ? it seems Visio Cafe is broken

[lusty jackal]

The download link works for me at https://www.visiocafe.com/netapp.htm

The link on the above page has the correct NetApp-Current-Index.zip, https://www.visiocafe.com/downloads/netapp/NetApp-Current-Index_2024-11-11.zip

[lusty jackal]

I am so happy that you can now join the cluster SVM to AD with "vserver active-directory create", no need for a domain-tunnel for cluster management auth. It's not in the What's New or Release Notes, but just tested it out in a lab and it works. FINALLY! Been wanting this for a decade

[stoic jolt]

hmm.. i just tried on another device and your correct.. I wonder if they were updating 🙂

[tawny basalt]

Um, @lusty jackal it’s really not new. It’s been around since like 8.1. Initially it was used when customers would purchase software without a cifs license

I think you still need the domain tunnel though.

https://kb.netapp.com/on-prem/ontap/da/NAS/NAS-KBs/How_to_configure_AD_authentication_for_cluster_when_CIFS_is_not_licensed

[stable grove]

He's talking about using the admin SVM. Before 9.16.1RC1 you would get this:

cl3::*> vserver active-directory create -vserver cl3 -account-name admin -domain test123

Error: command failed: This operation is supported only on a data Vserver.

With 9.16.1RC1 it's now possible to use the admin SVM and no, you don't need any domain-tunnel. Just tried it, works without issues via SSH and also System Manager.

cl5::*> vserver active-directory show
            Account         Domain/Workgroup
Vserver     Name            Name
----------- --------------- ----------------
cl5         CL5             LAB

cl5::*> security login domain-tunnel show
This table is currently empty.

cl5::*> security login show -vserver cl5 -authentication-method domain

Vserver: cl5
                                                                 Second
User/Group                 Authentication                 Acct   Authentication
Name           Application Method        Role Name        Locked Method
-------------- ----------- ------------- ---------------- ------ --------------
LAB\Administrator http     domain        admin            -      none
LAB\Administrator ontapi   domain        admin            -      none
LAB\Administrator ssh      domain        admin            -      none
3 entries were displayed.

cl5::*> security audit log show -vserver cl5 -timestamp >10m -username LAB\Administrator
Time                      Node         Audit Message
------------------------  -----------  -----------------------
[...]
Mon Dec 02 14:32:01 2024  cl5n1        [kern_audit:info:4094] 8003e800000f0873:8003e800000f0874 :: cl5:ssh :: [client-IP]:51290 :: cl5:LAB\Administrator :: Logging in :: Success
Mon Dec 02 14:36:47 2024  cl5n1        [kern_audit:info:10709] 8503e800000b7e71 :: cl5:http :: [client-IP]:51507 :: cl5:LAB\Administrator :: POST /security/login HTTP/1.1 : ["X-Dot-Client-App: SMv4"] :: Success: 200 OK

[olive umbra]

Yayyyyy

[tawny basalt]

Well with more context, that’s fantastic. Didn’t see any note about 9.16 until you posted it. Dealing with air gapped sites most of the time so I don’t encourage new code (still recommend 9.14.1p-latest until 9.16 hits p4). Therefore I don’t get too excited about the new features for a while

[atomic cargo]

hey idk if that's the right spot to ask but on a metrocluster IP config do we agree that you'd need to "send data" to both sites to use all disks right ? as half of disks are assigned to cluster of site A and half to the cluster of site B

[mossy ermine]

if you mean from a client point of view, then yes, a MetroCluster is basically two separate clusters so you have to divide your data between two clusters essentially. so you have some datastores on site A and some datastores on site B (which are each mirrored to the other side, respectively)

[atomic cargo]

ok ok ty just wanted to confirm that

[grave heron]

Is there any documentation from Netapp for setting up a netapp cluster as the backend for Grafana's Loki? I've read that Loki uses S3, do I need to have the S3 license on my netapp cluster?

[patent flame]

Drop that as a post in the #1062049169520476220 forum for me, please. Great question!

I'd also recommend looking into our OSS monitoring tool, Harvest, if you're looking for Grafana type solutions https://github.com/NetApp/harvest

[grave heron]

We use harvest for our fileserver monitoring, we're curious about using Loki for longterm compute/application monitoring

[patent flame]

Excellent, so you're already familiar. With our new ONTAP ONE licensing, there's no additional a la carte licensing anymore. The native ONTAP S3 isn't quite as feature rich as our StorageGRID product, but it has the basics.

[somber latch]

Hi All
Need a Power shell script to configure A150 NetApp storage

[mossy ermine]

what do you mean "configure"? day 0 setup? create SVMs? Volumes? LUNs? Replication?
you can use the PowerShell toolkit to do all that but that only includes the cmdlets, not completed scripts. If you specify what exactly you're looking for, someone in here might help but in the end you need to write the script yourself

[patent flame]

For day0 deployment, Ansible is going to be your best bet

PSTK is great for Day2 maintenance and mgmt

[mossy ermine]

yeah but if you only have to set up a single cluster for day0, doing that via ansible is probably overkill

[patent flame]

Fair

But there’s downloadable sample scripts ready to go. Just in case you’re already familiar with it

[whole wraith]

Hi guys, I’m not entireley sure if I’m in the right place or not however. I have a netapp appliance with one disk shelf and servers in the place of where the iom6 controllers would go. If I add another disk shelf to this and use qsfp cables to connect it does this then work? Thanks

[patent flame]

#┊・homelabs is what you’re looking for

[dark summit]

Sorry to bother you, but i need one quick info. Is there a way to alter the update sequence before issuing a ANDU in a 6 node cluster

[mossy ermine]

what do you mean by "alter the update sequence"?

you mean like which node gets updated first?

[dark summit]

jep