#╭・backup-and-dr

1 messages · Page 1 of 1 (latest)

woven spoke
#

Anyone doing MCC-IP in San Francisco Bay Area?

oak holly
woven spoke
#

Looking for others' experience with MCC-IP in the San Francisco Bay Area, such as what site pairs are far enough apart (in case of earthquakes) but close enough to meet MCC-IP's latency req of 7ms steady and <3m jitter.

fallen mountain
#

I've had past customers with DCs in SFO. They always had a 3rd replication site like DEN or ORD.

oak holly
#

it very much depends on the exact network type you use. 7ms brings you to about 500miles +/- a bit (700km).

#

lots of switches or bad switches might reduce it severly.

woven spoke
#

It sounds like MCC-IP may be still fairly new, at least to the SF Bay Area. How about our European friends, who I understand may be more familiar with MCC? What's been your experience with deploying across DCs that meet MCC-IP requirements of latency? What kind of distances between DR sites are realistic? How well does sync mirror keep up? What's the Switchover process like?

oak holly
muted dove
#

SnapCenter, Consistency Group snapshots and volumes on separate clusters. We have an Oracle database that is using volumes on two independent Netapp clusters. As far as I understand - Consistency Group snapshots can't span multiple clusters. Can SnapCenter work in this situation (without CG snapshots) and backup Oracle DB?

tired bobcat
#

No idea, but why do you have multiple clusters? Why not combine them then?

lucid ravine
#

Hello everyone! I have a quick SVM-DR question. I've noted that our SVM-DR config had the identity-preserve flag flipped to false by accident (likely during a DR exercise a while back) and that needs to be turned back on by recreating the relationship. What I'm wondering about is the impact to data and sync afterward? All of the volumes are present and would be fairly up to date (within an hour) upon break/recreation. Can someone give me an idea of what to expect?

fallen mountain
#

it'll be a resync.

lucid ravine
#

@fallen mountain Oh man - I actually looked at this earlier and didn't realize I needed to sign in.

#

It's Friday right? lol

pale quail
#

We’ve all been there

lucid ravine
#

Thanks for the help!

fallen mountain
#

No prob

oak holly
cunning terrace
#

The bigger question would be why do you need to hide some answers to kb-articles only for logged-in customers? Besides revealing the answer to fix or workaround an issue it's not like there are any more advantages to being logged-in on the KnowledgeBase. You can't interact or suggest any improvements etc.
I agree there are some cmds where it makes sense to only be revealed to employees and partners but for the most articles it just feels random. There is no consistency why this answer has been hidden and that one not.

#

Only my two cents

oak holly
#

you can very well interact. a feedback bar occur on the lower left after about 7 seconds (it occurs a bit delayed to not annoy too much if you hit a KB that is untreated to your issue)

cunning terrace
#

Yeah, but this also appears when not being logged in...

#

Also I just tapped on "No" and nothing happened 🤷🏻

Screenshot_20220806-125732.jpg
oak holly
#

it is provided in the backend. so your no is being looked at but without feedback basically not telling anything. I dont know why you dont see a comments box however that should be there

winter talon
#

@cunning terrace does other vendors open their KBs to everyone without authenticating?

cunning terrace
#

I don't know and honestly I don't care. This is about improving the NetApp KB and not comparing it to decisions of other vendors... Are you only hiding the answers because others do? Would you close down the KB if others decide to?...

The question is: What value do I get as a customer to being logged in the KB? And to turn it around: What harm is there to NetApp if the answers are visible when you're logged-out? (I'm not talking about deep dive cmds only visible for service-partners or employees)

Again I'm not against having an account and providing value but currently it's just confusing and many times people don't find an answer to their issue...

frozen nova
#

I’m glad @oak holly is part of the conversation here. He’s deeply involved with our knowledge management program and this is good feedback for the team that he and I are both part of.

oak holly
lucid ravine
#

@oak holly As mentioned, removing the login requirements would be best. At the moment half the article is already visible, which makes it worse as in this case it was enough to make me think I had the info. Glad to hear removing it is being considered. Thanks!

lucid ravine
knotty furnace
#

At least there is now a little pop-up that comes up when you go to a KB article that says "Sign-in Recommended" at the top. Never saw that before.

keen robin
#

I've been tripped up by that countless times. So that big "SSO isn't what you think it is, you have to sign in again" pop up is helpful.

winter talon
#

I agree it isn't a great experiance and it doesn't hold the login. I vote for if it's not internal we shouldn't force the login, cookies are enough

storm onyx
#

hi guys.
i just want create cluster peering using for snapmirror.

#

but my cluster still got pending

#

both intercluster is still alive.. in this situation what should i do ?

#

using NetApp Release 9.11.1: Tue Jul 12 10:21:46 UTC 2022

winter talon
#

Run cluster ping from both sides and see if you have any disconnects

#

You can also use Cloud Manager to automatically configure everything for you

storm onyx
#

still pending nw

winter talon
#

What is the result of cluster ping?

honest sluice
#

Anyone know how Hybrid Cloud Bundle is licensed for data that resides in a FabricPool? For the local tier it’s the amount of physical blocks you want to use SnapMirror Cloud for, but I have yet to see how the blocks that have been moved to StorageGrid counts. It would make sense if they also need licensing, but can’t really get that confirmed. And I have tried other Netapp channels. 🧐 Would also be very interested in hearing from anyone that uses SM-C in prod.🙂

fallen mountain
#

That's an internal URL my friend 🙂

frozen nova
#

Yep -- deleted

#

sorry @tired bobcat 🙃

fallen mountain
#

There's no cost to tier to StorageGRID (or ONTAP S3).

#

"but I have yet to see how the blocks that have been moved to StorageGrid counts." it's what data that gets moved to the remote tier(s).

fallen mountain
tired bobcat
#

Ooh oops.

fallen mountain
#

But there have been some changes to the Hybrid Cloud bundle if you're trying to quote it now

honest sluice
fallen mountain
#

oh.

#

not enough coffee yet

#

apparently

tired bobcat
#

What kind of system? I don't know if it matters FAS vs AFF.

fallen mountain
#

it don't

tired bobcat
#

Sounds like no one really knows?

#

😦

fallen mountain
#

trying to find the doc, but @sterile path you know off hand?

sterile path
#

SnapMirror Cloud is orchestrated by Cloud Backup Service and is licensed by the source capacity, pre-efficiency. Be careful if using ActiveIQ to find capacity as that is in TiB and sticker price is 50TB/month. As of this week, we can offer up to 50% off on Cloud Backup Service. Currently the licensing does not differentiate between StorageGRID as the object destination of Blob, GCS, or S3, but that could change in the future.

tired bobcat
#

"Currently the licensing does not differentiate between StorageGRID as the object destination of Blob, GCS, or S3, but that could change in the future." What about the source?

honest sluice
#

Potentially this customer is going to use Rubrik to orchestrate.

sterile path
#

No the source just has to be ONTAP running 9.7P5 or newer (but you should run 9.8+ for better experience) but CBS doesn't differentiate btwn source

#

there are 3 3rd party tools that can leverage the SM-C APIs w the hybrid software bundle and Rubrik was not part of that three. I am checking to see if that changed.

honest sluice
sterile path
#

Ahh I see! I messaged my friends in TLV to see about the SM-C w rubrik

honest sluice
#

But the question remains, how is Hybrid Cloud Bundle (and SM-C) licensed for tiered blocks, if at all.

sterile path
#

I asked that too 🙂

cunning terrace
#

I don't think having your source volumes on a FabricPool aggregate changes anything. ONTAP would just need to read the tiered blocks from the cloud tier (StorageGRID in that case) to back them up via SM-C orchestrated by Rubrik. If I remember correctly you license the logical used capacity so it should not really matter if it's all in a "normal" aggr, in a local tier or in the cloud tier. Efficiency is also irrelevant.

#

I think it says "Front-End Terabytes" somewhere

cunning terrace
#

Are both clusters on ONTAP 9.11.1?

honest sluice
cunning terrace
#

Oh wow, I didn't know that! Even though we managed to sell CBS already, this was one of the reasons it was unattractive for our customers

#

You pay for the S3 storage (the backups there maintain the original efficiency) but then you pay again for all the logical used capacity of the data you want to back up and here you lose all the benefits gained by ONTAP efficiency features ...

storm onyx
west hatch
#

📢 SnapCenter 4.7 has been published to the NetApp Support Site

bitter haven
#

I'm having a problem with the REST API and Snapmirror... It seems that the 9.11 (FSx) API doesn't support MirrorAllSnapshots policy.
"message": "Associating the specified SnapMirror policy with this SnapMirror relationship is not supported.",
"code": "13303866"
This works from the CLI, just not the API. I've searched through the KB site, with no solution found. Any suggestions?

prime terrace
fallen ether
#

You should be able to create your own API-based policy that behaves the same as MirrorAllSnapshots. You can do this today in System Manager.
In System Manager, Protection->overview->local policy settings->protection policies->+Add->Scroll down and check "Copy all source Snapshot copies"

bitter haven
# fallen ether You should be able to create your own API-based policy that behaves the same as ...

As it turns out, my solution was to create the SM relationship with the default policy, then make a call to the /private/cli API call to modify the snapmirror policy to 'MirrorAllSnapshots' . I believe adding a rule to an existing policy of type 'mirror-vault' might have worked, but unfortunately the SM policy modify functionality is unavailable on FSxN (policies belong to the cluster SVM, for which the command set is more limited than normal ONTAP).

fallen ether
#

Interesting. You can't really make a mirror-vault work like MirrorAllSnapshots, since the latter is an async-mirror type and designed for DR only (not backups). But glad you have a workaround.

north island
#

is SMB and NFS Multichannel still unsafe for backups and file transfer?

winter talon
#

Why not using block level with SnapMirror or Cloud Backup?

north island
#

Cloud is a NoGo.. and we use Veeam to backup so it incremental anyways. Im talking about the session layer protocol

winter talon
#

Both are on-prem, SnapMirror license is prob already included with your ONTAP

north island
#

oh ok NM_peepoShrugSmile its the first time i work with Netapp. Does SnapMirror and Cloud Backup need a second Netapp device for it? or does it work with every single server. What is the used protocol?

winter talon
#

SnapMirror is the protocol and it’s block level replication between NetApp arrays. Cloud Backup is a block level Backup between ONTAP and S3 compatible object store (StorageGRID, AWS S3, Azure Blob, Google GCS)

tired bobcat
north island
#

yea good question.. probably with the order of the packages i guess NM_peepoShrugSmile the post was pretty old tho

tired bobcat
north island
#

on a forum... but nvm... SMB3.0.2 cant be that unsafe

tired bobcat
#

Odd. Sounds old, but I have no context to go on.

#

Do you remember what it said?

north island
#

tbh i dont. It was a week ago.. it was also probably a early version on a QNAP

#

so its not representative for any device

tired bobcat
#

QNAP?

#

Yeah that's not anything we'd worry about. It probably was a QNAP specific bug anyway,

north island
#

as usual

tired bobcat
#

It's a bit like saying we're worried about a bug on Windows that affects graphics drivers but we run Linux and it has nothing to do with how Linux and graphic drivers work.

storm onyx
#

guys please share snapmirror transfer schedule best practise

#

Which policy would be appropriate to do it periodically so that there is no impact on the system? I am using MirrorAnd Vault now.

storm onyx
#

i can't use StrictSync because our volume is flexgroup vol!

pliant void
pliant void
crystal hornet
#

SnapCenter 4.7 released with new enhancements for Oracle, VMware, and Object backup targets!

rapid arch
#

Does new Snapcenter support file/folder level restores from flexvol?

west hatch
rapid arch
winter talon
#

It's more complex than you think 🙂

buoyant gate
#

Hi all! I got a problem with SVM-DR from 9.8 to 9.10
Missing SnapShot schedule on destination.
On earlier releases, I used to define the same schedule on destination cluster, but on 9.10 I get: "Error: command failed: Cannot create a schedule with the same name as an existing schedule in the cluster scope."
I see cron show has replicated schedules for SVM-DR relationships from another 9.10 cluster
How can I remideate this?

buoyant gate
#

Update, I got it running.
Seems 9.10 creates schedules on the source veservers when protected by SVM-DR. These are replicated to destination.
We have a SVM-DR source on 9.8, and this need schedules on the destination cluster.
I made new schedules on 9.10 source. After resync I was then able to create the original schedule name on destination cluster and SVM-DR from 9.8 was successful.
Not the most logical process..

cunning terrace
#

Just to confirm: Both your source and destination are the same ONTAP version, right? SVM-DR is not version-flexible.

#

But yeah, that's a new feature with ONTAP 9.10.1 that custom job schedules are replicated to the SVM-DR destination

buoyant gate
#

I have two source clusters. One on 98, one on 910.
Destination for both are 910

cunning terrace
#

ok, but the 9.8 to 9.10 SVM-DR relationship is not supported.

#

"Version-independence is not supported for SVM replication."

#

since SVM-DR is also replicating the config of the SVM you need the same ONTAP-version on source- and destination-cluster

#

it can work sometimes and for migration use-cases ("move" a SVM from cluster-a to cluster-b) I've also used it before, but for constant DR use cases it's not supported

buoyant gate
#

Are you sure?
From https://kb.netapp.com/Advice_and_Troubleshooting/Data_Protection_and_Security/SnapMirror/How_to_check_SnapMirror_compatibility_between_ONTAP_versions#SVM_Disaster_Recovery_(SVM-DR)
"SVM-DR source clusters are supported up-to two major ONTAP releases earlier than the destination cluster ONTAP version."

NetApp Knowledge Base
How to check SnapMirror compatibility between ONTAP versions

Due to a previous ONTAP Release model, SnapMirror source or destination clusters utilizing ONTAP 9.2 or ONTAP 9.4 may have limited compatibility with other major ONTAP Releases. Similarly, cloud-only …

proud herald
#

Here is a Massive warning to all people/customers who use TSM to do NetApp backups using SnapDiff.. As of 9.10.1 NetApp has disable/removed SnapDiff V1 and V2 and only has V3.. IBM has not signed on\licensed V3 will not be updating their software to support it. So if you upgrade your backups will break..

#

We have quite a few customers using TSM and backing up petabytes using snapdiff.. so i guess they will be on 9.9.1

#

one of our customers had and issue and NetApp support said upgrade and they just upgraded to 9.11.1 argh..

fallen mountain
#

@buoyant gate and @cunning terrace you're both right technically.
For DR. they need to be the same version for failover / fail back.
If you're doing migration it's 2 versions back.

#

I'll submit a ticket to see if the wording can be more clear in the KB and doc

buoyant gate
#

Thank you Mike. That makes sense. I'm doing migration, didn't think of failback here 🙂

#

Replaceing a 8040, highest ontap 98

shy pagoda
#

Hi All,
I am trying on the SVM DR.
With the SnapMirror relationship "type-XDP", the destination volume may not match the source volume total size.
Is there any way that I can disable this function, which is the same as "type-DP", when the next snapmirror scheduled, the source volume modify value will be reflected on the destination.

quartz citrus
proud herald
#

anyone seen this before with SVM-DR on the DR site sis.chkpoint.restore.failed: SIS operation on the volume '/vol/xxxxxxxxxxxx@vserver:0cd1c7fd-5583-11ec-bdfa-d039ea319ef1' failed to restore from a previous checkpoint. Starting SIS operation from the beginning.

#

getting a whole heap of these in the event log..

#

we are on 9.9.1P10.. there appears to be a BURT but it look like it fixed by our patch rev

subtle orbit
#

Hey @proud herald we've got an internal KB related to this particular bug, I would suggest getting a case open with Support as there is a couple of things they can review and suggest to help here.

buoyant gate
#

Is this bug only for DP-type mirrors, or also XDP?

#

(I was thinking of the problems with deswissler, not dedup. So my last question is probably not relevant. Sorry)

buoyant gate
#

I successfully migrated the customer in my questiones from 9/26 yesterday. Used SVM-DR on 9.10. Had to recreate a bit of config for iSCSI. NFS and SMB came up first try.
(For iSCSI, had to change initiator and LUN IDs, and recreate iSCSI LIFs, igroups and mapping)

mint bane
#

Hi NetApp help, i have a question:

We have multiple volumes on on our FAS8200 (A) that are being snapmirror to a 2nd FAS8200 (B) for backup purposes. On site A 2 snapshots are held and snapmirrored to site B where 14 restore points are held in snaplock mode.

Whenever i delete a backup file on the FAS8200 from site A the deleted data is being taken into the snapshot and snapmirror to site B, resulting in very large snapshots (Approx 10-15TB) each.

As a result the site B FAS is exceeding the limits and going low on storage space.
Because of the snaplock we can't delete any of the snapshots on site B to solve the issue.

Is there any way we can delete data on site A without it going in the snapshot and also being snapmirror to site B. I was thinking of queiescing the snapmirror relationship, am in on the right track here? any assistance would be kindly appreciated.

cunning terrace
#

Quiescing only means "pause the scheduled replication" so that won't help.
Snapshotting is always for the whole volume. If you want to preserve a snapshot from before you deleted the data it will always include that data. I mean... that's the whole point. Snapshots sort of "freeze" all the existing blocks of your active filesystem at one certain point of time so you can go back to it and restore data from that time.

#

Do you have SnapLock Compliance or Enterprise configured? With Enterprise you could delete the unneeded data from your SnapLock volume before the retention ends.

mint bane
#

Hi OG1,

Yes, so that means the data will still be transferred to the secondary site.. I don't want to preserve a snapshot however..
What i want is to clean up the volumes on the A site without the deleted data being snapshotted and transferred to site B and im trying to find a way to make that work.

Yes we have a 7 day snaplock compliance active on the destination volumes (vault)

buoyant gate
#

Are you able to controll when the files are created and deleted? If you adjust the snapvault updates so that the files you dont want to keep is already deleted before the update, they will not be transferred to site B at all.
Otherwise you are asking how to break the compliance.
If you know in advance what data you do not want to keep, you could also move those to another volume.

cunning terrace
#

ONTAP does not know which files you want to keep and which you don't. Snapshots will always enable you to recover deleted files as long as they are present when a snapshot is being created.
If you pause the snapshot-creation (for example with the snapshot-policy), then create a file, delete it and create a snapshot again, it will not be in any snapshots. This won't work if you want to delete an older file which already existed while at least one snapshot has been created.
Snapshots are always read-only and immutable. Yes, you can delete snapshots (if they are not on a SnapLock volume) but you can't alter the contents (for example delete a file which it references).

#

You could create a clone of the volume, delete the files you want there and then split the clone. This will create a new volume which needs the same space as the old volume minus the deleted files. Then you could activate snapshot creation again and delete the old volume.
The issue here is that you have a snapmirror-destination which needs a common base-snapshot with the source volume. Without this base-snapshot you can't use "snapmirror resync" to reestablish the relationship between the new volume and your destination volume.

astral sable
#

#╭・backup-and-dr Hi everyone,
I wanted to update you that Cloud Backup's Price Promotion Now Became the Official List Price.

You probably already know that we have the best technology to protect ONTAP with NetApp Cloud Backup. Now, it is also the most competitive and easiest to calculate, with the new volume price break.

For the last year, we had a promotional discount that helped us to win any deal. Now, this promotion becomes the official list price of Cloud Backup to make sure you can win the opportunities you face to deliver a complete solution of primary storage (AFF/FAS), secondary storage (StorageGRID) and tertiary storage leveraging the most efficient backup technology in the market.

The prices get as low as $0.017 per GB/Month, depending on customers’ protected capacity and terms of commitment.

You can contact me or @winter talon for further info, or check out this asset in the fieldportal: https://fieldportal.netapp.com/content/1925508****

ocean fern
#

Hi all,
I'm looking to update a 2 node Metrocluster we have setup. Can anyone advise on the behavior when done through the system manager (non-disruptive)?

fallen mountain
#

Haven't done one in a long while, but it's a switchover/switchback for a stretch MCC.
for the ANDU - The 4 node works well. haven't tried it on a 2 node though.

ocean fern
#

Yes, that's exactly the path I followed 😁 . So essentially, it's press the button and the MC will figure it out and switchover/back accordingly?

fallen mountain
#

yeah, run though the normal upgrade advisor, upload the image review the pre checks.

#

the 4 node was pretty seamless, just no switchover. it was TOGB in accordance with DR pairs.

#

what ontap version to what ontap version?

ocean fern
#

Thank you, that's cleared it up a bit.
We're looking to go from 9.7 -> 9.10 currently

fallen mountain
#

I think you might have to have a stop over on 9.8

#

9.11.1 has the 3 version skip option. i.e. 9.8 to 9.11.1

ocean fern
#

The upgrade plan mentioned that. I Just need to upload 9.8 & 10 and it should do it all in one go if i'm not mistaken?

fallen mountain
#

personally, i've not ever done the ANDU multi hop. just due to IMT reasons.

ocean fern
#

We'll see how brave we feel when we come round to it. Thanks for your help 😁

fallen mountain
#

no prob

cunning terrace
#

Is Fieldportal down at the moment?

#

I'm always getting:
We are unable to provide access at this time
A problem has occurred. Please contact your Netapp administrator for assistance.

frozen nova
#

Same. I'll get that looked at. Thanks.

subtle orbit
#

Yup, the squirrels that keep it working have gone on strike.

buoyant gate
winter talon
#

Hi @buoyant gate, the orchestration is done by Cloud Backup/Cloud Manager, and the data movement itself is done by ONTAP. ONTAP to StorageGRID is supported.
On pricing it’s probably a miscommunication or issue in the TCO calculation as it’s one of the most cost effective solutions there is, especially with the new list prices. Check out this TCO calculator https://cloud.netapp.com/cloud-backup-service-tco-calculator

Cloud Backup TCO Calculator

A simple way to estimate your cost to back up ONTAP data to the cloud using Cloud Backup. This service can be used for both on-premises and cloud-based ONTAP systems.

buoyant gate
#

Hi, @winter talon !
The question was on doing backup to local StorageGrid.
They are used to SnapVault, where there is no capasity licene, but if I understand correctly they will have to pay if they go to SG. One of my customers think this should be cheeper that going to a cloud.
Same as for FabricPool.

winter talon
#

I can relate to that everyone prefers to get things for free. It’s human nature.
Indeed I was referring to backup to local StorageGRID. At the end of the day it’s always a question if it’s required and what are the alternatives. Cloud Backup has the best data mover in the market, it simply backup/restore operations, it has next-gen catalog, and it’s priced lower than the competitors.

#

P.S. SnapVault isn’t free either

#

If you need any commercial assistance, you are welcome reaching out offline

buoyant gate
#

Any plans to support SVM-DR to a metrocluster?
For moving SVMs from non-metro to metro.
I plan on recreating SVMs on metro and mirror data-volumes onto that SVM. Is there a better way to do it?

south monolith
fallen mountain
fallen mountain
#

@buoyant gate work with your account team to submit and fPVR for the feature.

pliant void
shy pagoda
#

Hi All,
I am trying on the SVM DR.
When user resize the source volume, the destination volume size may not changed to the same size. (From XDP)
Is there any API that can resize both the source and destination volume size for SVM DR at the same time? or resize one by one ?

subtle orbit
eager rapids
#

Is there a way to backup ontap configuraiton backup to an HTTPS S3-Bucket from a StorageGRID? Currently we receive following Error Message: Error: command failed: Upload operation of configuration backup file exited with error: Failed to upload configuration backup file "/mroot/etc/backups/config/ConfigBackupNcchzrh96.7z" to"https://XXXXX/backup/AWSAccessKeyId=3ZPLINXXXXXX". Error: HTTP response code said error : The requested URL returned error: 405 Method Not Allowed.

tired bobcat
#

^answered on Communities I believe.

#

SG doesn't support HTTPS for ONTAP config backups but only HTTP.

brave junco
#

how can I restore a file from a cloud object store to a volume? I've created a SM relationship between the volume (destination) and the object store (source) with relationship type RST but the restore is failing with this error:

Error: command failed: "snapmirror create" is not supported for restore relationships between these endpoints.

fallen ether
brave junco
#

this is on behalf of a customer and this is the way they want to do it..

fallen ether
#

Is it FSx?

brave junco
#

yep

fallen ether
#

OK, well, the reason is because you don't create RST relationships, you simply run 'snapmirror restore' and the RST relationship is created by ONTAP. It gets removed when the restore completes successfully.

brave junco
#

thanks Joey! I'll have them give that a try

fallen ether
#

But...

#

It's not that easy. There are two things you need: endpoint UUID and the inode number. Manual single-file restores don't take file names to find the files in the obj-store, it uses inode numbers.

#

::> snapmirror restore -source-path <sm_obj_store_name>:/objstore/<container_or_bucket_name> -destination-path <restore_svm>:<restore_volume> -source-endpoint-uuid <endpoint_uuid> -source-snapshot <snapshot_name> -file-list <inode_number>,@/<dest_file_path_and_name>

brave junco
#

inode numbers to restore?! that's a new one for me

#

I guess I'm hoping something in the API layer is smart enough to do the filename -> inode translation for me

fallen ether
#

REST does it in two steps. With REST you actually do create the restore relationship first, then kick off a transfer. But yes, File_ID is still required.
POST /api/snapmirror/relationships : {"source.uuid":“<obj_store_endpoint_uuid>","source.path":"<sm_obj_store_config_name>:/objstore/<container_or_bucketID>","destination.path":"<dst_SVM:dst_vol>","restore":true} POST /api/snapmirror/relationships/<relationship_ID>/transfers : {"source_snapshot":"<obj_store_snapshot>","files.source_path":["<file_ID>"],"files.destination_path":["<destination_path>"]}

brave junco
#

in that case, file_ID is the inode number right?

fallen ether
#

yes

#

Of course, if you must do it manually, you could always restore the whole volume and pluck out whatever files you want afterwards.

#

Kind of overkill, but finding inode number can be tricky, which is why I started by asking about your backup application / OCCM. Unless I'm mistaken for every restore OCCM uses the the Active Data Connector to mount the cloud-side filesystem and get details like this about the file(s) being restored. I admit I don't know too much about how exactly AWS accomplishes this with FSx, but I assume it's similar.

brave junco
#

thank you! that's all very helpful. to do it from the CLI, I only need that single snapmirror restore command? nothing else?

fallen ether
#

If you have everything to plug into the command, yes that's all you need. You can monitor progress with 'snapmirror show'. If it works it'll go idle/healthy:true and then get cleaned up. If it fails it'll go idle/healthy:false and sit there until you fix whatever went wrong and try again or clean it up.

brave junco
#

thanks Joey! last question: how do we figure out the inode number?

#

back in the day I remember using i2p but that doesn't seem relevant here

#

it's not volume file show-inode because that shows a file for a given inode.. whereas I want the reverse

fallen ether
#

Well, that's the trick; as far as I know there is no command to do this from ONTAP (list an objects file inode that resides in cloud). As mentioned, OCCM does this with the Active Data Connector. I am not sure what's available for FSx, though I assume the same thing may be available I am not sure how you invoke it, per se.

brave junco
#

from one of Justin Parisi's old blog entries, set -priv advanced ; showfh shows the inode of a regular file in a regular volume (albeit it shows it in hex)

fallen ether
#

Those are ONTAP-side commands, they won't be helpful finding an inode number in a remote cloud object-store.

brave junco
#

hmm, it looks like we don't expose -source-endpoint-uuid in snapmirror restore so this might all be moot!

fallen ether
#

The "-source-endpoint-uuid" is the "-destination-endpoint-uuid" of the backup relationship which backed up the data originally. Assuming the backup relationship exists, you can get that with
set adv
snapmirror show -fields destination-endpoint-uuid

#

Or maybe you can't, I forget the restrictions on FSx with respect to changing command privilege. It's also in AutoSupport, if that helps at all, but it's a hidden field normally.

brave junco
#

oh - I mean it doesn't even accept the parameter, let alone the value.. maybe we don't expose it to customers

fallen ether
#

Oh I see

brave junco
#

I'll take a look. thanks Joey! this was a huge help

fallen ether
#

Well, I shot a question over to Engineering and they got back: "SFR (Single File Restore) is currently not supported in FSx because of lack of support for getting the inode number from a catalog service. This is an open requirement currently from FSx. BTW, user’s in FSx cannot use SM-C commands."

winter talon
#

Thanks Joey, I was about to comment that it’s not currently possible for FSxN, nor customers are allowed to use it.

#

Having said that, file level recovery it’s something we are looking at for FSxN/AWS Backup

#

Cloud Backup for ONTAP /Cloud Volumes ONTAP has two different technologies to recover files from the object-store directly to ONTAP volume:

  1. We are using an ad-hoc container which can scan the backup metadata and build a browsable catalog. (In just few seconds)
  2. Google like search which is build on top of Athena. (The fastest and cheapest catalog in the market)
    Both are completely transparent to the user and everything is automated according the restore best practices
#

BTW The Google like search is using SnapDiff v3 to track all of the changes

quartz citrus
fallen ether
quartz citrus
fallen ether
#

In system manager / GUI it's an option checkbox, I believe. But it translates to a special label used in the snapmirror policy rule.

quartz citrus
#

But you are saying it's just an otherwize empty rule with Snapmirror label of "all_source_snapshots" ?

fallen ether
#

Perhaps the docs need an edit for that

#

Every rule also contains a keep value. For all_source_snapshots the keep value is always 1, but that number doesn't reflect how the rule works and doesn't constrain what Snapshots are replicated. The other fields of the rule are default.

quartz citrus
#

Oh ok, thanks!
Just noticed I couldn't see the check box in the GUI because it's only showing when creating a policy, not editing.

fallen ether
#

Yea, editing a policy to add that rule doesn't really make much sense. all_source_snapshots is essentially the whole policy, so if you want to use that you're changing the policy itself rather than adding this rule.

quartz citrus
fallen ether
#

Just realized we are looking at the docs for SnapCenter plugin for vSphere, so that's probably why system manager looks different.

#

Right, it wants a policy purely for DR and wants to replicate all the snapshots

quartz citrus
#

I am in System Manager for Ontap, not the Snapcenter plugin

#

Thank you very much for the assistance. 👍

proud herald
#

is there any talk of MS adding snapmirror support to ANF.. AWS support snapmirror into FSXn ?

winter talon
#

FSxN does support SnapMirror

#

You can use BlueXP to manage these replications

brave junco
#

speaking of SnapMirror, I was yesterday years old when I discovered that SnapMirror supports TLS encryption! (I left NetApp around the cDOT 9.3 days)

south monolith
#

Along with >16T LUNs

quartz citrus
south monolith
astral sable
#

Now you can back up large-scale NAS data hosted on FlexGroup volumes
with NetApp BlueXP backup and recovery (Cloud Backup).

A first-of-its-kind innovation from NetApp

Read more here, and please share it:
https://bluexp.netapp.com/blog/cbs-blg-petabyte-backup-for-flexgroup-with-bluexp-backup-and-recovery

Petabyte Backup for FlexGroup with BlueXP Backup and Recovery

Petabyte backup has always been a challenge, until now. BlueXP backup and recovery can now help you protect the petabyte-scale data in FlexGroup volumes.

bronze burrow
#

The documentation for reversing a SnapMirror relationship (https://kb.netapp.com/Advice_and_Troubleshooting/Data_Protection_and_Security/SnapMirror/Guidelines_for_SnapMirror_Disaster_Recovery_(DR)_testing_and_reverse_SnapMirror_in_ONTAP) seems to be missing an argument:

e. Reverse the relationship again to return the original DR back to DR and the original Prod back to Prod
DR::> snapmirror -source-path svm_Blue:vol_1 -destination-path svm_Red:blue_vol_1_dr
                ^ ~~~~ what should go here?
NetApp Knowledge Base
Guidelines for SnapMirror Disaster Recovery (DR) testing and revers...

This KB is a walkthrough for a basic SnapMirror Disaster Recovery (DR) Test and reversal in clustered Data ONTAP and ONTAP 9. The procedure of this KB can be also used for SVM-DR. This KB can also …

#

resync?

pliant void
gaunt reef
winter talon
#

Would you see a value if Cloud Backup would manage the entire 3:2:1 solution?
For example: Configuring snapshots on the primary, set schedule and replicate to secondary, take local snapshots on secondary, and create backups to object store

crystal hornet
#

Ah! A great use for a poll!

ruby rainBOT
#

**:bar_chart: Would you see a value if Cloud Backup would manage the entire 3:2:1 solution? **

🇦 Yes
🇧 No
🇨 N/A

wooden crow
#

Is there a way to see/estimate the size of the initial SnapMirror base transfer?

Was looking around and can only find really old posts (2010/2012era)

fallen ether
#

If you're transferring all the snapshots (like with the MirrorAllSnapshots policy), the used size in 'vol show' against the source will tell you approximately how much you're replicating.
Using a vault or mirror-vault snapmirror policy will transfer just the source active file system (not the Snapshots) which you can find with 'df' on the source.

wooden crow
#

Sorry for educating me.. so SnapMirror doesn't just transfer the snapshots?

If I look at,
Vol snapshot show -vol ABC

And add up all of the sizes of the snapshots, I get about 12GB , but that volume is 2TB in size using 1.59 TB , physical 1.4TB of space, not much deduplication going on this Vol.

Is an initial SnapMirror going to need to transfer the 1.4ish TB or only the size of all the snapshots of 12GB?

fallen ether
#

SnapMirror does transfer Snapshots, but on the initial baseline it has to transfer every block the Snapshot points to which is the entire filesystem. After that, it only transfers the changed blocks.

#

The first transfer will be around 1.4TB.

wooden crow
#

Got it, thanks Joey!!

winter talon
#

Hi #╭・backup-and-dr ! We want to implement an integration with SIEM/DL/NOC/Ticketing systems and we want your help to determine our priorities for these integrations. Which tools are you or your customers are using??? (e.g, ServiceNow, Splunk, Slack, ELK, Tivoli, AIQUM/CI/etc...)

south monolith
crystal hornet
#

If you start with ServiceNow, Splunk, and maybe an SDK where people can build their own integration, that’d capture a large majority of the market

subtle orbit
#

From the support side of the world I see a lot of small to medium sized customers still just relying on Unified Manager. They're not at the scale where something like ServiceNow/Splunk is needed so UM + Email (with a little bit of Digital Advisor) is common.

tranquil aurora
#

+1 for an SDK/restful API for roll your own, having a framework to build off of and with can really help clients decide to make the leap. I know when I worked in healthcare expandability was king and was always something we looked for in tools, even if 'enabling' the feature cost a premium.

cunning terrace
#

ok I'll be the first 😩
You know about these Broadcom BES-53248 switches for your MetroCluster-IP? I just stumpled upon these shiny new "system switch ethernet log" commands in ONTAP 9.8. Do I understand it right that you will only need to / should use this to add your Broadcoms if they are Cluster-Backend switches? And not if they are Metrocluster switches?

#

It's only mentioned in the Broadcom setup guide (https://docs.netapp.com/platstor/topic/com.netapp.doc.hw-sw-ix8-setup/GUID-6703FC78-8D3D-47BB-8FA5-D1F036796585.html?cp=18_0_3_5) but I can't find anything about this in the Metro-IP config guide (https://docs.netapp.com/us-en/ontap-metrocluster/install-ip/index.html)

Configuring the cluster switch log collection feature

The cluster switch health monitor log collection feature is used to collect switch-related log files in ONTAP. You must make sure that you have set up your environment using the BES-53248 cluster switch CLI as detailed here.

MetroCluster IP Installation and Configuration Guide

Describes how to install and configure the MetroCluster hardware and software components in an IP configuration.

copper spade
#

If I understand your question, the setup is for cluster switch log collection configuration not for the Metrocluster switch log collection.

tidal badge
#

Yeah, that's the intended use, but would it work if you were also using those switches for MC-IP?

#

My expectation is you'd only need to do that at the request of Support, so probably not worth wasting anymore brain cells. Unless you're a TSE. 😂

mortal citrus
cunning terrace
#

Yeah, that's what I thought. I was just wondering because with Metro-FC you were always adding the bridges and FC-switches so ONTAP could monitor them (remember those StorageBridgeUnreachable mails...)
But with Metro-IP it looks like ONTAP does not really want/need to monitor the IP switches... dafuq

frozen nova
#

@pastel spoke I've moved your question over here:

Hello.. I have a quick question on volume snapmirror.. does setting "identity preserve" to true on volume snapmirror do anything? I know for SVM replication it copies all the config and permissions..

pastel spoke
cunning terrace
#

Both volume snapmirror and svm-dr is done with the same "snapmirror create" command, so the parameters should be the same... and for simple volume snapmirrors it shouldn't be set and not relevant

#

https://docs.netapp.com/ontap-9/topic/com.netapp.doc.dot-cm-cmpr-980/snapmirror__create.html?cp=5_2_21_2

[-identity-preserve {true|false}] - Identity Preserve Vserver DR
Specifies whether or not the identity of the source Vserver is replicated to the destination Vserver of the Vserver SnapMirror relationship that will be created. If this parameter is set to true, the source Vserver's configuration will additionally be replicated to the destination. If the parameter is set to false, then only the source Vserver's volumes and RBAC configuration are replicated to the destination. **This parameter is applicable only for SnapMirror relationships with Vserver endpoints. **The default value is false.

snapmirror create

Create a new SnapMirror relationship

pastel spoke
proud herald
#

any out there brave enough to use Veeam for NDMP backups of a Cluster?

#

the more the dig the more painful it seems it is

#

😦

nova rapids
#

Yep. I do it. It sucks. Don’t do it. I’m in the process of changing over to Commvault.

proud herald
#

ha.. ok

#

This is one of those ones where the client wants Veeam as its simple.. but looking into their support of NDMP its pretty dreadful and we have to do full volume backups as we are using Fabric Pool. So we cannot do file level backups

#

Veeam does not have API/Powershell for NDMP..

tiny pond
#

I have an OnTap 9.4 cluster (FAS 2650 & FAS 2554) that is replicating to a 2nd unit. I would like to preserve the snapshots on the replication target for a longer period. I get the message "Cannot configure Snapshot copies on this volume as it is a DP volume." Is there another approach I can use to accomplish this?

crystal hornet
#

Snapmirror is going to replicate/mirror the snapshots from destination to source. Personally, I would look at SnapVault as the alternative there, as you can set protection policies around how long to keep things.

tiny pond
crystal hornet
#

SnapVault does an initial baseline, similar to SnapMirror, but keeps incremental deltas above and beyond what the baseline sets. It's more of an incremental backup approach that a replication/mirror situation.

tiny pond
proud herald
#

Im about to SVM-DR a production SVM and DR from an older almost out of maint clusters to a new clusters. I have everything setup and its replicating across. Im just waiting to get the DR site final cabling up and running and i will cut that over first. Is there any DOCS/TR about what to do after you have cutover to clean up the snapmirror parts ?

#

also on that note its currently setup with a mirror vault from Prod to DR.. I assume after i SVM-DR the DR SVM and PROD SVM to new clusters will they automatically resume their mirrors ?

orchid marsh
orchid marsh
humble patio
#

I am trying to configure a demo environment using Lab on demand to demonstrate SnapVault + SnapLock for ransomware protection, but I am facing a lot of issues.
ONTAP 9.9.1 has no option on system manager to configure snaplock aggregate and volumes neither. I did everything using command line.
Using system manager is not possible to configure SnapVault to snaplock volume.
Does anyone could give me some tips ?

orchid marsh
#

its command line driven

#

System Manager features are not implemented yet as far I am aware

humble patio
orchid marsh
#

scheduled snapshots on the destination or scheduled transfers ?

humble patio
#

schedule the transfers... I was able the create the first transfer copy, but after that the transfers don´t occurs automatically

tired bobcat
#

Did you set up a policy for scheduling the transfers?

humble patio
#

yes... usign job schedule command and then I used this job in snapmirror command

tired bobcat
#

I think you apply the schedule to the policy, then apply the policy to the SM relationship.

orchid marsh
#

yea they changed that in 9.9 moving the schedule from the relationship to the policy but you should be able to apply it on either policy or relationship. it causes an error if you try to apply it on booth if the other has it already

proud herald
#

I have just done this for a client..

#

it was a PITA and not simple to get up and running

midnight vale
#

Is there a way to exclude volumes from SVM DR before the replication is established? You can only set vserver-dr-protection once the SVM DR is in place but it would be useful to choose which volumes to replicate before the first replication starts

orchid marsh
#

well that should work as soon the source SVM is peered with an DR Destination Vserver and created the SVM-DR Realtionship (uninitialized)

unknown.png
cinder pebble
#

Hello everyone, does anyone know a technical document for SQL server 2014 backup to StorageGrid?

orchid marsh
cinder pebble
#

I am using automated backup template written in SQL and created as a Job.

crystal hornet
#

It’s all GETs and PUTs, so if you’ve got a job that’ll write to S3, it’ll write to StorageGRID, just change the path where the job puts the file and you should be good.

orchid marsh
#

yea but the article is tailored around a AWS S3 bucket i did not see an option to switch to a SGrid bucket without including a 3rd party backup software

autumn vine
tired bobcat
tired bobcat
cinder pebble
crystal hornet
#

I need to get Pat (PapaSQL) in here

cinder pebble
tired bobcat
#

SQL

frail adder
#

Hello everybody!!

#

Do anyone use poweredge? I need some help please

tired bobcat
#

Maybe ask then we'll answer if we do. 🙂

mystic geyser
cinder pebble
rich elk
#

Has anyone used ElasticSearch onprem on ONTAP?

tired bobcat
#

On prem?

rich elk
#

Yep

tired bobcat
#

I thought that was an AWS service?

rich elk
#

It is a nosql db; it can be installed everywhere, but I'm trying to figure out how to best do it on-prem

tired bobcat
#

Oh. Well...what do you mean exactly by "figure out how to best do it"?

rich elk
#

ONTAP Best Practices... I'm going to use iSCSI, because they think NFS is too slow...

crystal hornet
#

The good news is, it's extensible data that could be pulled into a 3rd party engine

#

There are ..... "plans." 🤐

rich elk
#

Thank you. Unfortunately I need to get this setup on 50 hosts with 20TB iSCSI luns each; then back it up and maybe SM...

crystal hornet
#

So, in a nutshell, the way I like to think about it is ONTAP indexes at the block level based on things like aged date, access, etc. Things typically done in OS's at the file level. File System Analytics was implemented in ONTAP 9 for a variety of reasons to keep some of those attributes at the file level but we haven't taken it as far as fully globally indexed search capabilities yet.

tired bobcat
#

Honestly, that's a pretty open ended question.

#

I'd say your account team can help provide guidance and reach out to others as well.

#

Do you even have a filer yet?

#

(or are you looking to see what to purchase)

pale quail
#

Sizing elastic will be fairly similar to other apps like Splunk. I know there is some existing knowledge in doing this out there.

glacial yarrow
#

We have elastic running on VMware with NFS data stores as the backend. Runs no problems.

proud crescent
hollow owl
#

Hello there, would love to get some tips on my task.
We have on-premise s3 StorageGRID used for our apps, and we need to copy pretty large bucket (shy of 2TB, 4.6m objects) from one tenant to another. We also have no help on this task from team supporting SG deployment.
So far i've tried using rclone, it works fine as far as i can tell, but i have an issues with it - as i understand it does GET-PUT combo per file, so routes traffic through client and by my estimation based on test run it will take around 40-60 hours to complete the copy which is acceptable but
I was wondering if there are other tools or ways i can do the copy faster and more efficient.
I also saw this doc https://docs.netapp.com/sgws-114/index.jsp?topic=%2Fcom.netapp.doc.sg-s3%2FGUID-FAD01C83-41CE-4544-85ED-192BAFF9DEBF.html&resultof="put" "object"
and as far as i understand it may be used to copy content internally within SG with no routing through client, but without examples i have very little understanding how to use this functionality.
Thanks

hollow owl
crystal hornet
#

Hey @hollow owl let me see if I can get one of the storagegrid folks in here...

#

I believe there's a better way

hollow owl
#

Much appreciated, thx

crystal hornet
#

Are you on 11.4 or?

tired bobcat
#

What is the backend SG infra? Is it a VM or is it hardware like SG5700?

#

When you say move from one tenant to another, do you mean like vol move, or what?

#

Or is this just straight up SG and no FabricPool?

crystal hornet
#

I'm taking away that it's straight SG and moving buckets between tenants.

tired bobcat
#

Ah.

crystal hornet
#

Me and SteveP worked on something like this a couple years ago and it didn't require any kind of client.

tired bobcat
#

Yeah depends on backend too.

hollow owl
#

I have very little info about SG deployment sadly, I'm in a position of client

#

I've been given 2 pairs of ID-secret key for access, and Im positive that these accounts(tenants?) share same namespace for buckets

#

No idea about backend details

tired bobcat
#

Do you know the hostname or serial # of the SG appliance?

hollow owl
#

Interface is s3

tired bobcat
#

We can see if it's sending AutoSupport logs and see the details.

hollow owl
tired bobcat
#

Ok. Maybe get that info and see if someone who has access can help participate in this discussion. It may help to have them open a case too if it's under active support.

hollow owl
#

Okay, thanks. Will be back in a couple of days if I can get info.

crystal hornet
#

Hey @tidal badge ... got your ears on?

tidal badge
#

Yeah, checking now.

crystal hornet
#

@hollow owl meet Dan. Dan is our resident SG engineer.

tidal badge
#

Mikkie, recommendation from my team is to use s5cmd, it should be significantly faster

hollow owl
#

Thanks, i'll give it a try tommorow!

tidal badge
#

gl, let me know how it goes!

hollow owl
# tidal badge gl, let me know how it goes!

Done some testing - s5cmd operates much faster for sure, but it lacks ability to move data between tenants/accounts as far as i can tell. Will be my first choice if i need to move/copy between buckets on same tenant for sure.
I've had a talk with our SG deployment guys, they agreed to open case/ticket with support so we gonna try that.
Thank you all for help, was a pleasure.

crystal hornet
#

Thanks Mikkie, keep us posted and let us know if you need a nudge in support!

rugged tusk
#

Hello all, I've been tasked with restoring a backup from a physicial sql server to a virtual vmware sql server

The physical Microsoft SQL server cluster has fiber channel LUNs
The virtual SQL server has no direct FC connection to SAN switches

I have 3 options:

  1. Install Snapcenter plugin on the virtual server so we can try using snapcenter database restore feature (this is what I'd like to use)
  2. Use traditional SQL backup/export database from physical SQL cluster to bak files and restore to virtual
  3. Directly connect cloned FC LUN from physical cluster to ESXi host in VM environment so I can raw map the LUN to VM

Could Option 1 work?
Can you point me to a guide that discusses any of the options please
I've never done this before but will give it a shot

Thanks and Happy Friday!

tired bobcat
#

Maybe 1 might work. 3 might work too. Honestly I'd create a clone of the volume, make it a test LUN, and try it out.

rugged tusk
#

hmm, I didn't think of that. let me give that a shot. Thanks!!!

tired bobcat
#

Yw

crystal hornet
#

Personally, I would make sure I took a SQL bak of each mdb before doing anything. It's the safest but arguably longest way. Clone the new SQL VM and kick one of those restores off while you're mucking about with the other ways

rugged tusk
#

The SQL server is backed up nightly, so I'm safe there. I'm trying Paul's idea, in system manager I see the 2/6 snapshot I want. I did a clone of the volume. In my junction path , I mounted it and associated it with the default export policy. We have a vmware export policy but I noticed the svm I'm using doesn't have the nfs protcol (settings) configured.

proud herald
#

Hi All, In system manager when you create a new relationship you have the option for Volume, SVM or LUN.. If i have a single volume with a LUN in it do you choose Volume ? I assume LUN is when you have a volume with multiple LUNS and want to split out say just 1 LUN to a new volume ?

cunning terrace
#

In the background a relationship is always on volume or SVM level. I'm not 100% sure what the difference would be if you chose LUN in System Manager. It will still create a volume-relationship of the volume which includes the selected LUN/LUNs.

proud herald
burnt root
#

am a newbie in Netapp
Plz guide me how to start and how to deploy Net app storage in my desktop lab environment.

fallen mountain
tidal badge
hollow owl
#

Had a talk with techsupport at NetApp, they confirmed that there's no built-in replication functionality between tenants

barren iris
#

Hey friends! Question regarding CIFS, NFS, and possibly replication. Apologies if this doesn't belong here - I didn't want to hijack the discussion in #╰・software

Say there's a thin-provisioned NFS volume that was exported to a Linux server, and this Linux server has stored about 14TB data in this volume. Because of various customer-related shenanigans we need to do the following:

  • replicate that volume over to a different site (one that happens to be local to me)
  • set up a CIFS share so that a Windows jump VM at my site can access the contents of that volume
  • use a physical workstation that has access to this jump VM and copy the contents of that volume to an external hard drive

Predictably there are some permissions problems. I was able to complete the replication, break the snapmirror relationship, mount the volume, and set up a CIFS share. I can even see the contents (i.e., a bunch of folders) of the volume from the Windows VM.

However I can't access any of the folders - Windows tells me You do not have permission to access \\cifs-lif-ip\volume-name\folder-name. This error persists whether I leave the volume's security style as unix, change it to mixed, or change it to ntfs.

How fix?

tired bobcat
#

Can you use WSL2?

#

Could always use Windows NFS stack.

#

But my guess is either you don't have the usermapping, default user set, or something like that.

barren iris
#

Hm, both of those are new to me if I'm being honest. I might have latitude to Install Things on this jump VM or not

#

If you have some links to light reading I'd love to see them

tacit oar
#

@barren iris I believe you prob want the nfs volume in mix mode so that both nfs and cifs can access the files. it can be a mess to keep up changes from a windows user to unix

barren iris
tacit oar
#

its been a while so i dont remember if you need to or not

#

at the dest side you could also flex clone the nfs volume and that vol change the access type to cifs so that you can copy the files over

#

but really...setting up nfs client on windows is prob the easiest route to get this done

barren iris
#

yeah, gotcha, ty. I'll explore that

tired bobcat
#

You have to ensure user mapping works right.

#

If it's W10 or W11, honestly using WSL2 probably is easiest.

#

Or use Windows bulit in NFS client. Doesn't perform as fast but it is usable.

barren iris
#

Hm. Windows Server 2019 for what it's worth. The final workstation that'll transfer files to the external hard drive is likely W10

tired bobcat
#

Can you use Win10 to access the NFS share?

#

If so you're probably golden.

barren iris
#

Well, any data we're talking about would have to traverse this Server 2019 jump host before landing at its final destination

#

I'll explore WSL2 and NFS Client on this jump VM and see what shakes loose. I might even be able to get something installed. (Gotta love RBAC)

#

Thanks friends, appreciate it

tired bobcat
#

Yw

fallen mountain
#

go with a multiprotocol config

storm onyx
#

Hi Team..
Can configuration a single volume using for SnapMirror to multiple devices?

subtle orbit
crystal hornet
#

Yes, we refer to it as a “fan out” configuration.

coral fern
#

Hello all, Anyone ever had the message "Symptom: Following error observed on AIQUM Task Details Name: ndmp-connect(x.x.x.x,x.x.x.x) Description: Establish NDMP session with source X.X.X.X using login name admin and destination X.X.X.X using login name admin. Failure Reason: The NDMP service on cluster node cescnetapp2-01 is not enabled. Please enable it and try the restore operation again." This service, to my knowledge was never enabled. The day it ocurred i was out of office and nobody else touches the Netapp. I have not tried to do a restore so I am unsure why this error occurred. It appeared on April 6th and there appears to be no other errors regarding this.

tired bobcat
#

Is that node sending AutoSupports?

coral fern
#

No. It is on a air gapped network.

tired bobcat
#

Dang.

#

Well that's odd.

coral fern
#

I thought so too.

#

I am running a VMWARE environment from the aff and fas we have, I got SnapCenter and VSC runnig as well. But there are no errors with those.

tired bobcat
#

I would go into the audit logs from SPI and see where it might have come from

#

Maybe that might give you a clue.

coral fern
#

Dead end. Nothing about a restore, and if I understand the reading on NDMP, this si used for Tape backups, We don't have tape backups at all. This is a very curious error.

sour summit
#

To be clear, this is a question on what UM's alert is about, not about DR in and of itself.

Ancillary information because I already typed it and realized it was not accurate to the question asked...
There are a few code paths in UM concerning what is done on the cluster to restore something.

My suggestion: Check audit logs on UM, the cluster, any others integrated with ONTAP or UM. A user or another application orchestrating backups through UM, such as CommVault, may have triggered a restore.

NetApp Knowledge Base
OnCommand Unified Manager fails to restore file with error "Invali...
#

oh, even more accurately, this is not an alert, but in the details of a task. Even more-so, audit log reviews are the next step. Starting in UM around the timestamp of that task entry.

storm onyx
#

Hi team.

Can SnapMirror configure device to device DAS configure?
For example, FAS2720 e0f(10g) ↔️ FAS2720 e0f(10g)

I don't why customer wants to DAS configure he want to check about this.
i can't find this content on TR-4015 document so asked here.

Please Let me KNOW 🙂

fallen mountain
#

so between SVM1 and SVM2 on the same cluster?

storm onyx
#

no the other cluster.

#

change the example FAS2720 <-> FAS2750
the diffrent type of hardware.

fallen mountain
#

There's a inter opp chart for DP and XDP types.

storm onyx
#

Does it mean that 1:1 Direct Attached configuration between heterogeneous devices also doesn't matter?

fallen mountain
#

Oh. I see now why you said DAS. you mean just hard link a single 10 G cable.

#

ummm. I don't see why it wouldn't.

storm onyx
#

Right, the interface that the customer bought is 10G SFP type.

Customers want to configure SnapMirror by connecting 1:1 cables to different devices.

I asked because I was curious because I thought it would not be a problem because it would be possible to configure it with L2 communication is not recommended.

orchid marsh
#

yea I dont see why now DAS is normal a term for a server Direct attached Storage which kind of confuses with Snapmirror which is Transfer protocol on Network

storm onyx
#

Sorry, my English is not native, so I am not good at speaking lol

#

Anyway, if you can communicate L2 between different heterogeneous devices, there are no restrictions on configuring SnapMirror even if you configure it 1:1?

orchid marsh
#

well ahh wait direct then HA to HA makes a problem because one port would not reach booth HA of the other ones

#

that fails Full Mesh connectivity

fallen mountain
#

mmm yeah valid point. each icl needs to see each icl

#

I redact my should work statement.

storm onyx
#

i mean like this.

#

no L2/L3 SW just direct attach with heterogeneous devices.

orchid marsh
#

SnapMirror is using Intercluster network which requires that each node IC interface needs to reach each IC interface of the others cluster that KB describes that https://kb.netapp.com/Advice_and_Troubleshooting/Data_Protection_and_Security/SnapMirror/How_to_create_and_configure_Cluster_Peering_with_designated_InterCluster_LIFs_using_IPSpaces

NetApp Knowledge Base
How to create and configure Cluster Peering with designated InterCl...

This KB describes how IPSpaces are used to specify which InterCluster LIF(s) are used for cluster peering. The "full mesh" cluster peer requirement states all InterCluster LIF(s) in a …

#

so your DAS connection would be only supported Single Node cluster to Single node cluster

storm onyx
#

Thank you for your answer!

Then, in the case of a two-node cluster, it would be right to use a network switch.

It's not like you're going to connect a single controller to configure SnapMirror.

south monolith
#

you never know

#

🙂

regal sail
#

Is SM-BC working with a Hyper-V 2022 Cluster over FC ?

hazy oak
#

Hello, is it possible to convert a volume in snaplock volume ? I found a documentation with "vol copy" command, but the command doesn't exist again. Thank you

cunning terrace
#

What ONTAP version are you using?

hazy oak
#

9.10.1

fallen mountain
#

Saw this on the community :). but no it's not. I double checked the technical FAQ.

#

It needs to be copied out.

There was a blurb about using ndmpcopy, but i've not ever done that.

hazy oak
#

ok thank you

crystal hornet
#

You can SM into a SL'ed volume too, can't you?

#

I was doing that back in 2009

pale quail
#

I was about to say I think that’s possible, if there’s enough space, SnapMirror it into a new volume with the SL type set

fallen mountain
#

You can vault non-snaplock to snaplock.

#

Note: Although only mirror like-to-like relationships are supported with SnapMirror, vaulting non- SnapLock to SnapLock, is supported.

#

Is Jeannine or Tulledge on here?

frozen nova
#

I don't think so

hazy oak
#

thank you for this informations. Snaplock is not very flexible.

fallen mountain
#

There’s a lot of legal tied around it.

#

You can also work with your account team to submit an fPVR, worth a shot for enterprise I think because it is a little more forgiving than Compliance.

#

The Vault to Lock is also a good solution, and customers do run it.

crystal hornet
#

We were using SnapManager for Oracle to Vault and Mirror into SnapLock'ed volumes back in 2009, so I know this capability exists. SnapLock is build to NOT be flexible. The whole purpose of it is immutability to even admins, removing the ability for anyone to modify the data. I see it going around as a ransomware solution, but that's not really what it's intended for. It's meant for legal holds and long-term archival of any secure data (i.e. Healthcare has to keep patient records for 7+ years).

plush badger
#

Hello,

We have a snapmirror cascade and we would like to change it to a normal snapmirror, does anybody have a description with ontap command how to do it ?

current situation : A-> B -> C

source volume in cluster A has a snapmirror relation to Cluster B wich has a snapmirror relation to cluster C

Final situation : A -> C

we want to remove Cluster B to have the source volume in cluster A snapmirrored to cluster C

What are the steps and command line sequence to remove Cluster B from the cascade and to synchronize A with C ?

Thanks a lot for your help,

BR,
Vhyse

winter talon
#

You can you just need to resync from the new destination

#

SnapMirror resync

plush badger
#

thank you

proud herald
#

Has anyone used SMB-C with ONTAP ? If we have a dark fibre (layer 2) connection between 2 sites and that will be where the Intercluster LIFs are patched into and IP's assigned. Does the mediator software need to be on this Layer 2 network or does it the normal Admin Network for monitoring ?

cunning terrace
#

If I understand it correctly the connection between the Mediator and the clusters happens via the cluster-mgmt interface. By default it uses port 31784 with TCP.
I could not really find any mention of this in the SM-BC docs but at least the (internal) technical guide says: "SMBC uses a set of REST API services". Since you can only talk REST APIs with mgmt-LIFs that should be correct.
(with MCC-IP the Mediator also uses iSCSI but that is apparently not the case with SM-BC)

subtle orbit
opaque dragon
#

Hi. We are using a StorageGRID as FabricPool for our cold data. Hot data is stored on a netapp cluster. Is it possible to use SnapMirror for the hot data only? The cold data is allready mirrored the a second storageGRID using storageGRID functionlities. So that doesn't need to be mirrored again, only connected to the snapmirror destination. Can you give me a hint if this is possible and where I can get more Information about that task?

cunning terrace
#

What tiering policy are you using for your volumes inside the FabricPool?

#

You could use FabricPool mirror to tier cold data to two object stores simultaneously

#

But afaik using ILM replication and erasure coding for the cold data is also supported with StorageGRID

tired bobcat
#

I agree we need more data.

#

If you have two clusters with FabricPool and SG, and one is a SnapMirror source and the other is a destination, I believe you can set tiering policies on the destination different than the source.

cunning terrace
#

If I'm not mistaken a SM replication will always replicate both hot data in the local tier plus the cold data in the cloud tier.

#

It just depends if the destination volume is also on a FabricPool

#

on pages 13/14 there's a helpful table

opaque dragon
#

He @cunning terrace and @tired bobcat thank you for your reply. I am using Auto tiering policy on the Main Cluster to tier cold data to the storageGRID 1 after one month. The cold data gets than replicated via ILM Policy to another storageGRID 2 at a different site. I now want to use SnapMirror to replicate certain volumes. But, since the cold data is allready replicated, I now only want to replicate the hot data and map the cold data from storageGRID 2 if this is somehow possible.

tired bobcat
#

Oh. I think it would need to rehydrate because the metadata is different.

#

I don't think it is.

opaque dragon
#

So it would be better in this case to replicate both, cold and hot data via SnapMirror and tier it from the snapmirror destination to storageGRID 2 instead of using ILM? And thanks for the TR, i will Read that in depth.

cunning terrace
#

Hmmm.... Not sure what would happen if you add the bucket from SG 2 to the destination volume local tier

tired bobcat
#

Right from my understanding. I can ping a couple folks to confirm.

cunning terrace
#

But the TR says that you shouldn't use the same bucket for different clusters

tired bobcat
#

Oh there you go.

opaque dragon
#

Right okay. I will read it once I'm back in the office. Thank you :)

cunning terrace
#

Think about it this way: Using erase coding or even replication on your SG is mainly to increase the availability of the data inside the FabricPool aggr. It's not really about backup since you're always missing the hot data in your local tier and all the metadata stuff which is also local. The same with using FabricPool mirror. Always keep in mind that tiering means moving blocks from A to B and not **copying **them.
So you would always need to protect your volumes by creating another copy for example via SM. On your destination system you could then for example use the All tiering policy to tier even more data. But the cloud tier of your destination volumes shouldn't be on the same systems which also provide the buckets for the source volumes. Think about it, production data and backup data would be on the same storage which makes no sense.

#

But yeah it gets a bit complicated with big StorageGRIDs and replication.

junior plinth
#

Snapcenter: can I use the same transaction log drive for another instance in the same SQL cluster?

regal sail
#

Are there any plans to support SCSI-3 Persistent Reservations with SMBC, so that Hyper-V Cluster Luns are fully supported?

south monolith
#

Sorry!

crystal hornet
#

That said, reach out to your account rep and request a briefing or answers and they can connect you with the right prod mgr and get you that info. Hopefully you understand we just can’t discuss that stuff here publicly.

rich elk
#

Does anyone understand "partial snapshots" and their ramifications? I am trying to expand my flexgroup, but do not want to lose backups

fallen mountain
#

TLDR - can't do a snaprestore, but can still access files.

gritty horizon
#

I just saw this when setting up CBS on my CVO cluster. When hovering over a greyed out volume i see the following error message "Volume does not have any SnapMirror labels defined in source Snapmirror policy"

#

Any ideas?

sly fern
gritty horizon
#

Yes they are. Of the 30 volumes in the SVM, all of them have the same snapshot policy and 4 of them get the warning message above. I've opened a case so we shall hopefully get some movement soon.

#

I'm doing all of this from the SVM-DR target cluster which for me makes it difficult to see the underlying snapmirror config for each volume.