╭・open-source

[pallid cypress]

I am not sure where this would be placed, but here we go. How many people leverage a Unix based LDAP such as OpenLDAP or a bigger version of it.

What I am wanting to do, accomplish, for my testing now and in the future, is to spin up an basic OpenLDAP image\vm\etc. and be able to plug and play from that point forward.
I have tried following various articles and end up getting stuck somewhere along the lines. I am using plain old centos virtualbox, but would use ubuntu or anything else if it was easier and worked.

If anyone has a really generic image/iso/docker, etc I would love to get a copy.

What I have attempted but seems to fall short...

https://computingforgeeks.com/install-configure-openldap-server-centos/
https://www.cyberithub.com/best-steps-to-install-and-configure-openldap-server-on-rhel-centos-7-8/#Step_1_Prerequisites
https://www.ezeelogin.com/kb/article/how-to-install-openldap-and-phpldapadmin-in-centos-426.html
https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/linux_domain_identity_authentication_and_policy_guide/krb-nfs-client

I often use lab on demand for many tests.

[feral ridge]

Looks like Bitnami has an OpenLDAP container image and the README for the repo looks fairly comprehensive: https://hub.docker.com/r/bitnami/openldap/

^^ have you tried that one?

[pallid cypress]

That is what I found as well. It woked well after I figured out how to expose ports and how to get an env spun up in ldap with a user:pass. Apache Directory Studio was an easy tool to install to GUI manage my LDAP instance.

[stray falcon]

https://www.redhat.com/en/blog/red-hats-commitment-open-source-response-gitcentosorg-changes

[golden wasp]

Reddit comments are saying they’re all going to Debian ¯_(ツ)_/¯

[olive gate]

Sure they are. Unless they're enterprise customers in which case this mostly doesn't matter. When you have very expensive software licenses for software that only works or is supported on specific versions of RHEL, Debian is not an option.

[stray falcon]

Feels like a lot of blustering to me. I for one support the move to streamline operations and remove distractions.

[sage ingot]

There's going to be FUD spread by those who had expectations set, but at the end of the day Red Hat is still operating within GPL. I can see educational and HPC being impacted, but those that were dependent on CentOS (or derivatives such as Scientific Linux) have already made the decision to go to Stream or something else. I think the main issue most are taking is that Red Hat had said that the source code would remain on git.centos.org back in 2020, and some decided to make a downstream distro based on that directive. But, times change, and if your distro lives or dies based on a single repo that you do not have control over... time for a plan B.

Taken from: https://www.centos.org/distro-faq/

Devil's advocate; it'll be more difficult than it used to be to develop for RHEL without a support contract. Developer licenses do exist but this isn't a hurdle that is experienced with other distros/companies.

[stray falcon]

Well said, Renny. I fear this will do more market perception damage to IBM/RedHat in the long run than any particular product, or development thereof.

It also feels like there are some political chess moves happening behind the scenes here that aren't being shared publicly.

[potent bough]

I would love to see NetApp qualify and support additional Linux distros (non-RedHat clones, like Debian, SuSE, Ubuntu, whatever) for their Linux-based software (e.g. ONTAP Mediator, StorageGRID appliances, etc.)

[feral ridge]

I know of a number of folks that use Amazon Linux on-prem. AWS publishes AL2 vm and container images. You can get support for AL2 on-prem through 2025, at present. AL2023 is now out, but vm images for on-prem deployment have yet to be launched. Amazon Linux is based on the Fedora source, so unaffected by the RHEL announcement.

[stray falcon]

I always thought Fedora and CentOS (among others) were downstream “community” distros of RHEL. Hell, they even named the distro after the damned red “hat”

How is Fedora not affected?

[lofty wind]

The second sentence here indicates that Fedora is upstream from Red Hat Enterprise Linux https://docs.fedoraproject.org/en-US/quick-docs/fedora-and-red-hat-enterprise-linux/#relationship-between-fedora-and-red-hat-enterprise-linux

I personally use Fedora everywhere I can at home so I am often introduced to features and changes before observed at work.

[stray falcon]

https://opentf.org/announcement

Well, this seemed inevitable.

[deft flame]

It'll be interesting to follow this and see how it plays out the next couple of years. The BSL isn't completely horrible, especially for projects that are primarily supported by a single company. Terraform has enough interest though that OpenTF could be a better option depending on how well Hashicorp handles things from here.

[lament quartz]

Their silent update to terraform registry terms is interesting....

basically disallowing using anything other than hashi's terraform with it.

[deft flame]

Looks like the OpenTF people don't think this is a huge impact and creating another registry would be a long term goal anyway

[stray falcon]

https://youtu.be/TT_WbbK5u3c?si=-FAtCrg9AY2sazYo

[marble hazel]

Hi, i am wondering if this is the correct place to ask some general NAbox related questions ? like why one cluster i am trying to add, refuse to gett added- i managed to get it in the systems tab by manually adding it to yml config, but in grafana it doesnt count it so there is technically no poller created, and webgui if i click save, just ticks off green hook on all parameters, but then doesnt save the page, so it never goes through.

[golden wasp]

Hi @marble hazel - you're close 🙂
Head over to #1062050414146625536 for that

[marble hazel]

Ok, that means in general creating a post, and describing an issue following procedure? , or is it simple enough to say that the documentation is outdated, and you can figure the way yourself- and then if you want make a post about it you can. .. i mean im sure i would get to goal in less than 30seconds of actual chat with someone.
happy for the guidance, and again just wanted to clarify. as i see no means of writing directly in that channel

[golden wasp]

Those channels are what Discord refers to as Forums channels. It helps keeps conversations more organized when there are multiple topics being covered. It also helps the teams monitoring those specific channels to keep up with what's going on.

Just click "New Post" and go for it

[marble hazel]

Sure, but then i might aswell put it in the netapp support official pages, if i need to create a post. and i was looking for the quick quick solution, as i am on my free time atm.

Fully understand and respect the fact you aint got that, but then the point of me being here is dissapearing

[golden wasp]

The post is really no different than chatting here like we are now.

The devs watch #1062050414146625536 and not necessarily this channel.

[marble hazel]

Ok, well. To me it is a big difference, and thank you for the responses. Appreciate that.

[golden wasp]

Can you help me understand why you feel that way?

[marble hazel]

I am trying....the production is at a level of confidentiality, that it is already questionable why we would even consider running anything opensource related in direct communication with the storage nodes. i dont know how i can address the issue without either speaking to someone who knows what i am speaking about, and can give some quick pointers- or have to document in process a "post" where i dont really want even to be discussing it in public in the first place, sanitizing various texts and so on to even be allowed to talk.
so by that i dont feel like spending more of my personal time on creating tickets leading sometimes somewhere, other times nowhere. when all i need are some quick answers that the FAQ didnt cover weirdly enough. - i will probably have it sorted myself during the night of repeating/failing until it sticks. eliminating all, one by one..and that would still be sooner than the rate of any of the posts have been responded too in there.

But that it takes me less time to figure and deploy x node elasticsearch clusters with graylog in containerized environment production ready in less time than to get this up and running have been mindblowing for my patience. and im probably shit aswell. so Have a good day sir. i hope you understand now, . now just look at how much time that took to reply to.

[golden wasp]

Understood. Your original message asked where to post your question, and I provided that guidance. The message you wrote in the #1062050414146625536 channel can be as simple or as detailed as you're comfortable with. The message you provided above is fine as a starter. The NAbox team are marked as Staff and if you want to take a conversation to a DM with them, that's an option some choose to go with. If you want a quick answer, your best option is to ask over there.

[marble hazel]

I understand, and if I had more time, available that i felt i could do that. (i just keep doing changes and kicking the containers, rinse n repeat -> restarts of the docker hierarchy since we started chatting its been quite a few., at some point ill get the thingie correct and eliminate whatever bugs there are vs "following the documentation online". so i can just go on myself) and, yes naturally i dont like having that mindset or sounding overly negative- but at the end of the day, when the customers are making economical claims. and when the official support fails to provide any slight feasable theories to why that is, and now that i need to create a ticket for another system in an unsupported, inofficial forum to be able to look at problem/related official Netapp, its beyond me. It is a total disappointment and a way of pissing on a 20year+ old customer relationship who otherwise felt the products where stable, robust and giving you value for the expensive cost it is.

[golden wasp]

All I'm asking is for you to copy and paste your original message (#╭・open-source message) into a new "post" in the NAbox channel. Nothing more.

Discord is currently where we're supporting these open-source tools, and it's how we've done it for years. If you'd like to change this process, I'd encourage you to have the conversation with your account team. They'll be better equipped to help shape the future of support for open-source resources.

[marble hazel]

Ok, i will try.

On a principle level it doesnt matter to me if youre supporting it on discord or somewhere else, its more the fact that this started with some latency issues for a few (very few, select of delicate customers that share some infra on a 10 000+ customer machinery, and as being explained on the supportline that the system manager aswell as the netapp core product itself does not do inddividual measurements, but averaging so we have no way of telling wether the customer experience is real or not. and that official support wants us to resolve the problems regarding that on our own with open source solutions. ..,following that we have it up and running now, but there are bugs and unclarities and the documentation seems to be very outdated and so it either seems we will be unable to get the figures we need to pinpoint where the problem is or , as such disprove or prove netapps being the problem. but without an exact answer, the solution will be to migrate these customers off the netapps over to another vendor's bit for bit diskwise, the same system) where we did not have any of these issues, we are not forwarded to a community for our paid support and so on.

This is following a vendor assisted upgrade to 9.13.1P4 and we had none of the issues before. , we went from 9.8 following a step-by-step route to get there, we were intending to continue upgrading reminding nodes aswell but are afraid of continuing since these problems costs us quite alot in terms of revenue regressed.

[golden wasp]

That's a fair point and I appreciate having your perspective on this.
If you've got a ticket open to troubleshoot an issue and support's suggestion was to use these tools to help narrow it down, then they'll be able to connect the right people to the case and assist further. Let's see how things progress in the thread. https://discord.com/channels/855068651522490400/1200214236702441622
Thanks!

[thin adder]

is there any documentation on the netapp_ontapp python librarie's resources/classes that describe the properties and usage of methods? im afraid looking at the endpoints and their descriptions isnt helping me understand how to use the library...

[feral ridge]

Is this the one you're referring to? https://github.com/NetApp/ontap-rest-python

Did you ever get this question answered?

BTW, @golden wasp - the README for that repo still references The Pub Slack. 🙂

[golden wasp]

Thanks! I’ll get that updated

[thin adder]

i eventually stumbled across some blogs from people explaining how they ysed the specific parts of the api i needed, but it seems that the python library is only partially documented and if you want to go deeper in other parts of the library you need to look at the api documentation itself and not the python module

[feral ridge]

Yeah, the last time I looked at it, it felt more like a generic wrapper. Likely that was to make it lightweight so they were not having to do new releases with every version of ONTAP.

[subtle matrix]

Apache Cassandra 5.0 went GA at the start of this month. Fun fact - Did you know a couple of key NetApp Instaclustr staff committers were critical in getting this new major version to market? It's also now available on the NetApp Instaclustr platform.
Check out the release blog post here: https://www.instaclustr.com/blog/apache-cassandra-5-0-now-generally-available/

[timber quest]

Trident fails to create backend by 401 Unauthorized due to credentials problem Trident

No joy in making backend connection. Confirmed credentials with test api call with curl.

[oak elm]

Try it again (let it fail) then look at the event log on the command line of the Netapp
“Event log show”
It should tell you why it’s failing. You may be using or specifying the incorrect application/method

[gleaming junco]

Hi,
I am trying to build my own NetApp VM on my Linux remote server to perform a POC. I want to develop software that will be used to scan files into the volumes and detect vulnerabilities. I tried downloading Host Utilities from the support page but got an "Unauthorized access" issue. Is there a way that I could integrate it easily?

[potent bough]

what do you mean "your own NetApp VM"? NetApp is the company, and they sell multiple products. If you mean ONTAP (the main product) then you need the "ONTAP Simulator" which you should be able to download as soon as you have registered on their website (it might take up to 24 hours until you are allowed to download anything as the system is not fully automated).
But you can also do any development against any random NFS and/or SMB server, as those are the main protocols that ONTAP uses to share files. So if your tool can scan a CIFS share, it can also scan an ONTAP volume (same with NFS)

[random narwhal]

Maybe this would help?
https://www.flackbox.com/how-to-build-a-netapp-ontap-lab-course

[faint rune]

Don't know where else to ask this, but was directed to the "Developer Hub" channels:

Looking for a discussion forum about "VMware Aria Operations". Is there a PAK file that can be added for setting up the adapter to get metrics on our Netapps?

Seeing the ontap stuff mentioned here, but nowhere to find the PAK file: https://docs.netapp.com/us-en/netapp-solutions/vmware/netapp-aria.html

[faint rune]

Anyone?

[golden wasp]

Any of the <@&1010202775621226637>s know how to help out?

[thin steppe]

I think this is the closest channel for my question as I'm mostly looking ready-made and hopefully open-source product:

We have projects wanting to do book-keeping and cleaning of their allocated storage (volume or qtree) in filesystem level. Currently we're running ncdu/gdu to get the information needed but going through 10-40 TB and at worst tens of millions of files per project, tend to use resources quite a bit.
So question is, has someone seen products, or have been experiencing of getting similar in-depth information through NetApp's File System Analytics via REST? Adding non-admins to System Manager does not sound the wisest approach, even though granting only RO access.

[potent bough]

I know there are external quota tools like QFS from NTPSoftware that can probably help with that, but that's not open-source

[thin steppe]

Thanks for the info! Unfortunately after quick googling I came to believe that QFS product is long gone and anyway seemed to support mostly SMB/CIFS. We work with NFS shares, so might have not been optimal solution anyway.

[dire solar]

I apologize up front, but I'm really feeling like a moron right now. I have NAbox setup and I'd like our Prometheus instance to scrape the metrics. I am a total n00b with Prometheus/VictoriaMetrics. We have a Consul instance (?) that will scrape exporters on 9479. I'm trying to figure out what I need to put on that port. The node-exporter instance? I also apologize if I'm totally butchering all of these terms

[dire solar]

Is it possible to install Consul on NAbox?

[ebon jungle]

not sure, nabox uses flatcar linux. You would most likely have to compile it specifically for the system. However, i think the system is designed to not allow things like that, since it is 'secure by design' and immutable.
You should as in the #1062050414146625536 channel

[dire solar]

Posted a question in the #1062050414146625536 discussion/forum(?)

Is there just not much traffic here or there? Not complaining or anything, just trying to calibrate my expectations

[golden wasp]

Plenty of traffic in the harvest forum, as you’re probably seeing now 🙂

[dire solar]

(I did also send an email to the nabox help list with a link to my post 🙂 )

But I did get an answer and confirmed that it works

So I'm quite happy

[hollow nest]

Hello hello everyone! My first post here!

I am fairly new to NetApp (month number 4 here), but I have been working in Open Source software for 10+ years now (working in software waaay longer than that 😂) . I have been in the AI/ML ecosystem for several years now working as an AI/ML Engineer and Developer Advocate... if you are ever up for chatting about anything AI/ML, drop me a line!

In the meantime, if have 3 minutes to spare... Help shape the future of the NetApp Developer Hub! Take our short survey to share your voice! https://forms.cloud.microsoft/r/6wY5mJtKyx

I am apart of the effort to help improve Developer Hub and your feedback would be wonderful.

Cheers!

[golden wasp]

<@&1132032648282390538>
Your input on the Developer Hub is greatly appreciated!

[stray falcon]

https://community.netapp.com/t5/Tech-ONTAP-Blogs/Revisiting-Developer-Communities-for-the-Next-Generation-of-Builders/ba-p/462784

[slim turret]

Hi folks - I have a question about Cinder support for NetApp OnTap NFS. The OpenStack docs on netapp.io say that online volume extension is not supported. The upstream Cinder matrix has conflicting answers (but the latest version says no) and there’s been some chatter in the upstream cinder project itself about it. I can’t tell if it won’t or can’t be supported, or if it’s just not supported yet. If it can be supported, is there a timeframe?

I'm asking as I work in tech marketing/community management for Platform9, and our Private Cloud Director product uses best of breed OpenStack + other open source software. We've run into a number of NetApp customers that want to use NFS with Cinder and have asked about online extensions. Thanks

[stray falcon]

@torn meadow you around to assist?

[hollow nest]

If you are interested in why Enterprise implementation are starting to move over to use Graph for their RAG implementations, check this blog post: https://community.netapp.com/t5/Tech-ONTAP-Blogs/From-quot-Trust-Me-quot-to-quot-Prove-It-quot-Why-Enterprises-Need-Graph-RAG/ba-p/462813

[torn meadow]

Hello Damian, NFS extend for in use volume doesn't work. There was some issue with documentation, which we have fixed in the latest one but need to update that in older release.

Engg has a patch already prepared for this purpose. However, enabling this functionality also requires support from the Nova team. The timeline for implementation is contingent on the Nova team's progress. We are aiming to incorporate this enhancement in the upcoming G-release.

[slim turret]

Thank you!

[stray falcon]

Thanks Deepak!

[slim turret]

Hi Deepak - my company, Platform9, may be interested in picking this up in our Private Cloud Director product before mainstream OpenStack releases it, as we already patch upstream OpenStack quite a bit. Would you mind shooting me an email? damian at platform9.com

[dapper surge]

Hey everyone!
I just released an open-source two-level storage simulator that lets you test and compare placement and migration policies. It's fully customizable — you can configure each storage tier (e.g., NVMe, HDD, etc.) with any specs you want.
Check it out here: https://github.com/hocinemahni/mc_arc
Feedback and contributions are welcome!

[stray falcon]

Thanks for sharing! Will check it out!

[hollow nest]

New blog: DocumentRAG Using OpenSearch: GraphRAG-like Structure Without the Graph Overhead

GraphRAG has exploded in popularity for its structure, but it forces teams to maintain a full graph ontology and schema. This post introduces the BM25-based Document RAG Agent (or what I am calling DocumentRAG), a practical middle ground between VectorRAG and GraphRAG that preserves explainability while avoiding graph overhead.

Read the full blog post here: https://bit.ly/48kWHvC

[hollow nest]

Forgot to post the follow up to the blog above called: Hybrid RAG in the Real World: Graphs, BM25, and the End of Black-Box Retrieval

We discuss why Hybrid provides a 96% factual faithfulness on answers when compared to plain vector embeddings and we also provide an alternative to the Hybrid RAG (Graph + Vector) you typically see out with an alternative BM25 + Vector Hybrid RAG solution. This BM25 + Vector variation provides an in between solution that doesnt require the heavy lift of using a whole new database and maintaining graph ontologies/structure, while still getting most of the benefits that Graph portion provides: factual data grounding in answers.

Take a look at the blog post here: https://bit.ly/4pz0D3b

[hollow nest]

Dropping a new blog hot off the press titled:
Less Compute, More Impact: How Model Quantization Fuels the Next Wave of Agentic AI

Bigger models used to win headlines. Now they win (in not good ways) with power bills. This post looks at what changed after DeepSeek R1 made it clear that smarter engineering can compete with brute force. Instead of chasing parameter counts, we look at quantization, fine-tuning, and specialized Small Language Models that focus on one job and do it well. We also unpack what this means for agentic systems, where multiple focused models collaborate instead of one giant model trying to do everything.

This shift is happening for a reason. GPU costs are rising, data center power demand keeps climbing, and inference is now the line item that finance teams watch closely as token costs rise. NVIDIA’s recent inference-focused deal with Groq signals the same trend: latency, efficiency, and cost per token matter more than raw size. If you are building AI systems today, the question is no longer how big your model is. It is how much value it delivers per watt and per dollar.

Dive into the full article on the Open Data Science blog: https://bit.ly/4s6iKye

PS: for those interested in this topic, I will be presenting this topic at 3 different conferences (SCaLE this week, in April I have Devoxx France and a workshop at ODSC East) and 1 podcast due out Friday.

[hollow nest]

If you are interested in the blog post above on why Small Language Models and Quantization are going to going to see a dramatic uptick in Agentic solutions....

I recently joined the Open Data Science Conference (ODSC) AI X Podcast with Sheamus McGovern to talk about what's actually happening inside production AI systems. Not the polished demos. The messy reality is when models meet budgets, latency limits, and infrastructure constraints.

We covered a lot of ground in this conversation:
• Why many RAG and agentic AI demos fall apart in production
• The shift from bigger models to smarter-per-watt systems
• What quantization really does when you move from FP32 to INT8 or INT4
• Why Small Language Models (SLMs) often work better for multi-agent systems
• Hybrid RAG architectures that combine vector embeddings with knowledge graphs
• The growing need for governance and observability in enterprise AI

🎧 Listen to the podcast:
Spotify: https://bit.ly/40jPqsq
Apple Podcasts: https://apple.co/4bfx3tu
SoundCloud: https://bit.ly/4s0UrlV

[quasi rune]

This is a great listen!

[hollow nest]

thanks!

[hollow nest]

New Blog Post Alert:
Engineering Inference: KV Cache, Shared Storage, and the Economics of AI

Large language models burn through GPU memory and compute faster than most teams expect. Every prompt creates key-value tensors that sit in GPU memory, and that memory footprint grows with every token and every user. In this article, I walk through what is really happening inside KV cache systems and why architectures like vLLM and LMCache exist in the first place. Instead of treating caching as a performance trick, the post looks at it as a memory strategy that changes how inference systems are built.

This topic matters right now because the economics of AI are shifting. Training made the headlines, but inference is what drives ongoing cost in production systems. Techniques such as KV cache reuse, memory tiering, and shared storage are becoming critical for controlling GPU spend and data center power consumption. As companies deploy chat systems, RAG pipelines, and agent workflows at scale, engineering the inference stack is becoming more important than adding more GPUs.

Dive into the full article here: https://bit.ly/4bl87kn