<div dir="ltr"><div dir="ltr"><div>Hi, </div><div>first of all thanks to Thierry for driving this SIG creation.</div><div>Having a SIG to discuss how to deploy/operate a large deployment will be incredibly useful.</div><div><br></div><div>In my opinion we shouldn't restrict ourselves to a specific project or deployment size (or number of cells) but discuss the limits of each project architecture, the projects dependencies, limitations at scale (functionality vs simplicity), operational difficulties...</div><div>Sharing experiences and understand the different challenges and actions that we are using to mitigate them will be extremely valuable.</div><div><br></div><div>I think that we already have a lot of examples of companies/organizations that are deploying OpenStack at large scale. Compiling all this information (Summit presentations, blogs, superuser articles, ...) will be a good starting point for all operators and discussions. Every deployment is different. </div><div><br></div><div>I also would like this SIG to be the bridge between the operators of large deployments and developers. Bringing specific pain points to discussion with developers.</div><div><br></div><div>cheers,</div><div>Belmiro</div><div>CERN</div></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, Nov 14, 2019 at 8:25 AM Arnaud MORIN <<a href="mailto:arnaud.morin@gmail.com">arnaud.morin@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex"><div dir="auto">Hi all,<div dir="auto"><br></div><div dir="auto">+1 for me and my employer (OVH).</div><div dir="auto">We are mostly interested in sharing good practices when deploying a region at scale, and operating it.</div><div dir="auto"><br></div><div dir="auto">For the deployment part, my main pain point is about the configuration parameters I should use on different software (e.g. nova behind wsgi). </div><div dir="auto">The current doc is designed to deploy a small pod, but when we are going large, usually some of those params needs tuning. I'd like to identify them and eventually tag them to help other being aware that they are useful at large scale.</div><div dir="auto"><br></div><div dir="auto">About operating, I am pretty sure we can share some good advices as well. E.g., avoid restarting neutron agents in a single shot.</div><div dir="auto"><br></div><div dir="auto">So definitely interested in that group. Thanks for bringing that up.</div><div dir="auto"><br></div><div dir="auto">Cheers.</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">Le mer. 13 nov. 2019 à 19:00, Stig Telfer <<a href="mailto:stig.openstack@telfer.org" target="_blank">stig.openstack@telfer.org</a>> a écrit :<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-style:solid;border-left-color:rgb(204,204,204);padding-left:1ex">Hi Thierry & all - <br>
<br>
Thanks for your mail. I’m interested in joining this SIG. Among others, I’m interested in participating in discussions around these common problems:<br>
<br>
- golden signals for scaling bottlenecks (and what to do about them)<br>
- using Ansible at scale<br>
- strategies for simplifying OpenStack functionality in order to scale<br>
<br>
Cheers,<br>
Stig<br>
<br>
<br>
> On 13 Nov 2019, at 11:18, Thierry Carrez <<a href="mailto:thierry@openstack.org" rel="noreferrer" target="_blank">thierry@openstack.org</a>> wrote:<br>
> <br>
> Hi everyone,<br>
> <br>
> In Shanghai we held a forum session to gauge interest in a new SIG to specifically address cluster scaling issues. In the past we had several groups ("Large deployments", "Performance", LCOO...) but those efforts were arguably a bit too wide and those groups are now abandoned.<br>
> <br>
> My main goal here is to get large users directly involved in a domain where their expertise can best translate into improvements in the software. It's easy for such a group to go nowhere while trying to boil the ocean. To maximize its chances of success and make it sustainable, the group should have a narrow focus, and reasonable objectives.<br>
> <br>
> My personal idea for the group focus was to specifically address scaling issues within a single cluster: basically identify and address issues that prevent scaling a single cluster (or cell) past a number of nodes. By sharing analysis and experience, the group could identify common pain points that, once solved, would help raising that number.<br>
> <br>
> There was a lot of interest in that session[1], and it predictably exploded in lots of different directions, including some that are definitely past a single cluster (like making Neutron better support cells). I think it's fine: my initial proposal was more of a strawman. Active members of the group should really define what they collectively want to work on. And the SIG name should be picked to match that.<br>
> <br>
> I'd like to help getting that group off the ground and to a place where it can fly by itself, without needing external coordination. The first step would be to identify interested members and discuss group scope and objectives. Given the nature of the group (with interested members in Japan, Europe, Australia and the US) it will be hard to come up with a synchronous meeting time that will work for everyone, so let's try to hold that discussion over email.<br>
> <br>
> So to kick this off: if you are interested in that group, please reply to this email, introduce yourself and tell us what you would like the group scope and objectives to be, and what you can contribute to the group.<br>
> <br>
> Thanks!<br>
> <br>
> [1] <a href="https://etherpad.openstack.org/p/PVG-large-scale-SIG" rel="noreferrer noreferrer" target="_blank">https://etherpad.openstack.org/p/PVG-large-scale-SIG</a><br>
> <br>
> -- <br>
> Thierry Carrez (ttx)<br>
> <br>
<br>
<br>
</blockquote></div>
</blockquote></div>