<div dir="ltr"><div>Hi,</div><div><span class="gmail-im"><br></span>
> Hmm; you're talking about changing the host OS depending on what gets
scheduled to it? That'd make everything kind of loopy =) Normally one
instead schedules workloads to compatible hosts, requiring the hosts to
be prepared out of band.<span class="gmail-im"><br></span></div><div><span class="gmail-im"><br></span></div><div><span class="gmail-im">No. I am not saying to reconfigure a node every time you have a different container with a different set of requirements.<br></span></div><div><span class="gmail-im">What I am saying is that we will be working on a data source that can be used by different models to configure/provision worker nodes, and that could be used on a running cluster.<br></span></div><div><span class="gmail-im">It would be up to you to decide how and when you should feed Machine Config operator with this. It doesn't even have to be automatic.<br></span></div><div><span class="gmail-im">The data could be used to identify a specific group of applications with similar compatibility requirements (like enablement of SR-IOV) to configure a target group of nodes.<br></span></div><div><span class="gmail-im"><br></span></div><div><span class="gmail-im">> </span>What's the expected time commitment from the working group?</div><div><br></div><div>Join the meeting once a week and express your ideas/concerns.<br></div><div>How much time you would like to spend on tools implementation is up to you.</div><div>For now the most important thing is to collect feedback/opinions/ideas from the various stakeholders.<br></div><div><br></div><div>>
Let me see if I can find others at Red Hat who may be interested in this too.</div><div><br></div><div>Thanks.</div><div><br></div><div>Regards,</div><div>Marcin<br></div><div><span class="gmail-im"><br></span></div><div><span class="gmail-im"><br></span></div><div><span class="gmail-im"><br></span></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">czw., 21 wrz 2023 o 14:52 Colin Walters <<a href="mailto:walters@verbum.org">walters@verbum.org</a>> napisał(a):<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Hi,<br>
<br>
On Tue, Sep 5, 2023, at 9:26 AM, Marcin Franczyk wrote:<br>
> Hello,<br>
><br>
> I am writing to ask if at Red Hat you would be interested to <br>
> collaborate on container image standard improvements under Open <br>
> Container Initiative (<a href="https://opencontainers.org/" rel="noreferrer" target="_blank">https://opencontainers.org/</a>).<br>
><br>
> At Huawei we identified that container image specification lacks <br>
> compatibility requirements. <br>
> For instance if you run a container that requires an NVIDIA GPU then <br>
> you have to make sure that CUDA library in the container matches the <br>
> CUDA driver on the host. The required version of the driver could be <br>
> expressed in the container image. There are plenty of similar use cases <br>
> when it comes to kernel configuration, boot args, modules or <br>
> out-of-tree drivers.<br>
><br>
> More details could be found here: <br>
> <a href="https://docs.google.com/document/d/1lzwh8DGMu5vXXHwJmnewYIMffkcOEvH8owX4UYjRcw0/edit?usp=sharing" rel="noreferrer" target="_blank">https://docs.google.com/document/d/1lzwh8DGMu5vXXHwJmnewYIMffkcOEvH8owX4UYjRcw0/edit?usp=sharing</a><br>
<br>
I added a comment in <a href="https://docs.google.com/document/d/1lzwh8DGMu5vXXHwJmnewYIMffkcOEvH8owX4UYjRcw0/edit?disco=AAAA5kwp_40" rel="noreferrer" target="_blank">https://docs.google.com/document/d/1lzwh8DGMu5vXXHwJmnewYIMffkcOEvH8owX4UYjRcw0/edit?disco=AAAA5kwp_40</a><br>
><br>
> I think OpenShift and its Machine Config operator could benefit from <br>
> this by influencing worker node configuration based on the image <br>
> specification, which would mean that customers would also benefit.<br>
<br>
Hmm; you're talking about changing the host OS depending on what gets scheduled to it? That'd make everything kind of loopy =) Normally one instead schedules workloads to compatible hosts, requiring the hosts to be prepared out of band.<br>
<br>
> I am forming a working group under OCI and believe this initiative <br>
> should not come only from one company since that will be a global <br>
> standard.<br>
<br>
Right.<br>
<br>
><br>
> I prepared a template for a working group that will be voted on by the <br>
> Technical Oversight Board (<a href="https://github.com/opencontainers/tob" rel="noreferrer" target="_blank">https://github.com/opencontainers/tob</a>).<br>
> <a href="https://docs.google.com/document/d/1WZbr7xEpUohvyIiSvJEgCCxhAVMM9DmgXBsk1q1MtVU/edit?usp=sharing" rel="noreferrer" target="_blank">https://docs.google.com/document/d/1WZbr7xEpUohvyIiSvJEgCCxhAVMM9DmgXBsk1q1MtVU/edit?usp=sharing</a><br>
><br>
> The voting will take place once I create a PR.<br>
> I think, after the working group is approved, we would need 2-3 months <br>
> to come up with the first standard and a client tool to build and <br>
> validate the specification.<br>
<br>
What's the expected time commitment from the working group?<br>
<br>
Let me see if I can find others at Red Hat who may be interested in this too.<br>
_______________________________________________<br>
CentOS-devel mailing list<br>
<a href="mailto:CentOS-devel@centos.org" target="_blank">CentOS-devel@centos.org</a><br>
<a href="https://lists.centos.org/mailman/listinfo/centos-devel" rel="noreferrer" target="_blank">https://lists.centos.org/mailman/listinfo/centos-devel</a><br>
</blockquote></div>