ceph - sepia - 2025-01-16

Timestamp (UTC)Message
2025-01-16T04:09:55.523Z
<Mark Nelson (nhm)> did the vpn just go down for anyone else?
2025-01-16T04:11:33.825Z
<Mark Nelson (nhm)> just came back
2025-01-16T05:57:56.400Z
<nehaojha> @Dan Mick Ankush is around, can you create a bug tracker of some sort to summarize the issue for his team?
2025-01-16T06:02:10.110Z
<Ankush Behl> Hey @Dan Mick Sorry, I didn't got the notification of the mentions, I will try to fix it. Let me see what could be done to fix it.
2025-01-16T07:03:22.223Z
<Dan Mick> [www.ceph.io](http://www.ceph.io) is running eleventy to generate the static site.  It's very old and so are its dependencies.  We need a node.js/npm dev to update it
2025-01-16T08:40:16.374Z
<Ankush Behl> Okay
2025-01-16T08:40:34.890Z
<Ankush Behl> @nizamial09 @Afreen can you help with this?
2025-01-16T08:55:24.873Z
<nizamial09> Yup, i can have a look and try to update it
2025-01-16T09:25:15.866Z
<Matan Breizman> Because of the bugs we must require for gcc13/clang16. Seems that our builders are using clang so that's the main concern atm.

Beside the (untrivial) bugs which cause us to hack around them, Crimson's main submodule officially supports only the last two major releases of gcc/clang. It's probably ok for us to lag behind that for a while but latest clang stable is already at 19 today (e.i 5 major releases)
2025-01-16T09:26:48.372Z
<Matan Breizman> Because of the bugs Crimson's cmake must require for gcc13/clang16. Seems that our builders are using clang so that's the main concern atm.

Beside the (untrivial) bugs which cause us to hack around them, Crimson's main submodule officially supports only the last two major releases of gcc/clang. It's probably ok for us to lag behind that for a while but latest clang stable is already at 19 today (i.e 5 major releases)
2025-01-16T09:27:53.799Z
<Matan Breizman> Because of the code generation bugs, Crimson's cmake must require for gcc13/clang16. Seems that our builders are using clang so that's the main concern atm.

Beside the (untrivial) bugs which cause us to hack around them, Crimson's main submodule officially supports only the last two major releases of gcc/clang. It's probably ok for us to lag behind that for a while but latest clang stable is already at 19 today (i.e 5 major releases)
2025-01-16T09:29:12.425Z
<Matan Breizman> Because of the code generation bugs, Crimson's cmake must require for gcc13/clang16. Seems that our builders are using clang so that's the main concern atm.

Beside the (untrivial) bugs which cause us to hack around them, Crimson's main submodule officially supports only the last two major releases of gcc/clang. It's probably ok for us to lag behind that for a while but latest clang stable is already at 19 today (i.e we're 5 major releases)
2025-01-16T09:29:23.701Z
<Matan Breizman> Because of the code generation bugs, Crimson's cmake must require for gcc13/clang16. Seems that our builders are using clang so that's the main concern atm.

Beside the (untrivial) bugs which cause us to hack around them, Crimson's main submodule officially supports only the last two major releases of gcc/clang. It's probably ok for us to lag behind that for a while but latest clang stable is already at 19 today (i.e we're 5 major releases behind)
2025-01-16T11:53:30.124Z
<Ankush Behl> Thanks @nizamial09
2025-01-16T15:22:05.303Z
<Ronen Friedman> ... and we are missing full support of C++20(*). (C++23 was approved in 2023... just saying).
(*) no workable ranges, for example.
2025-01-16T16:46:17.336Z
<Yuval Lifshitz> i'm trying to force an ubuntu run, using `teuthology-suite -d ubuntu` but i still get a centos test.
any idea?
2025-01-16T16:47:27.342Z
<Casey Bodley> `-d` doesn't work well since we changed to yaml-driven distros
2025-01-16T16:48:28.325Z
<Casey Bodley> if the job you're scheduling uses a random distro, you can `--filter ubuntu`. otherwise you'd need to change the yaml on a suite-branch
2025-01-16T16:50:15.295Z
<Yuval Lifshitz> thanks! it works 🙂
2025-01-16T18:44:43.326Z
<Adam Kupczyk> Hi guys!
I have problem re-running failed teuthology tests:
```teuthology-suite -v --rerun akupczyk-2025-01-15_09:59:53-rados-aclamk-testing-nauvoo-2025-01-14-1834-distro-default-smithi --dry-run
2025-01-16 18:38:18,245.245 INFO:teuthology.suite:Using rerun seed=6486
2025-01-16 18:38:18,246.246 INFO:teuthology.suite:Using rerun subset=(382, 13264)
2025-01-16 18:38:18,246.246 INFO:teuthology.suite:Using rerun no_nested_subset=False
2025-01-16 18:38:18,249.249 INFO:teuthology.suite.run:Checking for expiration (None)
2025-01-16 18:38:18,249.249 INFO:teuthology.suite.run:kernel sha1: distro
2025-01-16 18:38:18,478.478 DEBUG:teuthology.repo_utils:git ls-remote <https://git.ceph.com/ceph-ci.git> main -> bd0160de81e216e42d835a3d4ce920c3bef81b16
2025-01-16 18:38:18,478.478 INFO:teuthology.suite.run:ceph sha1: bd0160de81e216e42d835a3d4ce920c3bef81b16
2025-01-16 18:38:18,478.478 DEBUG:teuthology.packaging:Querying <https://shaman.ceph.com/api/search?status=ready&project=ceph&flavor=default&distros=centos%2F9%2Fx86_64&sha1=bd0160de81e216e42d835a3d4ce920c3bef81b16>
2025-01-16 18:38:18,610.610 INFO:teuthology.suite.util:Container build incomplete
...```
and it is right, because the build is this:
<https://shaman.ceph.com/builds/ceph/aclamk-testing-nauvoo-2025-01-14-1834/feec94cd5a3db57edf0198d29ed3562f9f0ee11e/>

Why is teuthology-suite looking up for "bd0160de81e216e42d835a3d4ce920c3bef81b16" ?
2025-01-16T18:59:52.219Z
<rzarzynski> Hi @Adam Kraitman, [o08.front.sepia.ceph.com](http://o08.front.sepia.ceph.com) seems quite unreliable these days. It's unreachable while o06 is fine
2025-01-16T19:14:01.376Z
<Dan Mick> The infrastructure team is willing and able to help deploy fixes to the Ceph build code to make new compiler releases available, but does not have the expertise to be the authority on which compiler versions are deployed where, or how choices are made so as not to affect backward compatibilty with older Ceph releases.   that is to say, if someone has a plan, please implement it in ceph-build, do the testing, and let us know if you need help
2025-01-16T19:29:25.635Z
<Dan Mick> regarding update of shaman: it looks to me like the primary server is the one who must update the db.  The secondary can accept writes, but it forwards them to the primary for the actual write operation.  I don't know how long the cluster would operate if the primary were down (if the secondary would just buffer the writes, or if the data for other reads would be stale, or what).
2025-01-16T19:29:47.385Z
<Dan Mick> regarding update of shaman that we discussed in the infrastructure call this morning: it looks to me like the primary server is the one who must update the db.  The secondary can accept writes, but it forwards them to the primary for the actual write operation.  I don't know how long the cluster would operate if the primary were down (if the secondary would just buffer the writes, or if the data for other reads would be stale, or what).
2025-01-16T19:56:06.452Z
<David Galloway> AIUI, if 1.shaman goes down, 2.shaman is supposed to take over
2025-01-16T19:56:12.317Z
<David Galloway> Accepting writes
2025-01-16T19:57:40.444Z
<Dan Mick> I don't think so.  I think 2.shaman can accept writes, and does when it's 'load-balanced' by nginx on shaman, but it then forwards those writes to 1.shaman.  2.shaman has a strictly readonly access to the db.
2025-01-16T19:58:17.993Z
<Dan Mick> (as I updated in the wiki page)
2025-01-16T23:04:55.119Z
<Laura Flores> Hey @Dan Mick @Adam Kraitman what would it take to get an email list made for the Ceph website ([ceph.io](http://ceph.io))? David and I are now designated reviewers, and we want to add a note to the documentation that people outside the Ceph organization should contact us via email list to get their change approved. Here's the context: <https://github.com/ceph/ceph.io/pull/788>
2025-01-16T23:05:33.558Z
<Laura Flores> If you guys are not the right people to ask, I will look elsewhere- just not sure who actually creates the email lists.
2025-01-16T23:06:01.186Z
<Dan Mick> just some admin on [lists.ceph.io](http://lists.ceph.io)
2025-01-16T23:06:44.931Z
<Dan Mick> doc-review a good name?
2025-01-16T23:06:53.286Z
<Dan Mick> well not really docs
2025-01-16T23:07:00.455Z
<Laura Flores> ceph-website
2025-01-16T23:07:00.921Z
<Dan Mick> website-admins?
2025-01-16T23:07:09.327Z
<Laura Flores> That's what we called out GH team
2025-01-16T23:07:27.523Z
<Laura Flores> That way, it's a catch-all for anything related to the Ceph website
2025-01-16T23:07:58.578Z
<Dan Mick> okay.  I'd prefer a function in the alias name, but we can change later if need be
2025-01-16T23:08:41.516Z
<Laura Flores> ceph-website-admins is fine too
2025-01-16T23:10:09.238Z
<Dan Mick> created, you and David's [ibm.com](http://ibm.com) addresses added as owners
2025-01-16T23:10:20.979Z
<Laura Flores> Thanks Dan!
2025-01-16T23:10:24.290Z
<Laura Flores> Appreciate it

Any issue? please create an issue here and use the infra label.