For the meeting on Monday, 24 November 2025. S390/PPC Testing Upstream? ========================== Radoslaw noted a recent PR (https://github.com/ceph/ceph/pull/64894) that is S390/Power specific. We'd like to merge it, it looks reasonable. In general, we don't test S390 and PowerPC upstream. Since we're moving the lab, is this on our plate/plans at all? Greg points out that we don't just lack machines, we don't really have anything in place to support those architectures upstream. Perhaps we should, but we'd need to convince some corporation to support it and donate hardware. The answer for new architectures is always that we are happy to accept hardware, we will not make your github checks required, and you have to do the work. (Especially since many of them are very expensive.) Ilya says that IBM has some virtual machines that can be carved out. Several upstream projects are using that to test on S390. It's hardware made available by IBM in a cloud-like setting. Probably not sufficient for our needed testing. We might put a github check there on a low frequency, maybe a couple times a week on main. Someone needs to watch it and address its output for it to be any use. We would need continuous time investment to do the work. Casey thinks downstreams should set up nightly/weekly builds on their own hardware and fix things when they break. building/testing upstream puts the maintenance burden on us. Lab Shutdown - 15 December ========================== David Galloway reminds us that 15 December is when we will start shutting the Sepia lab down. A truck will be arriving two days later to pick everything up and take it tou Poughkeepsie. It is increasingly looking like we will not have access to the stuff in Poughkeepsie at the end of the month as we were told. So it is looking increasingly like there might just be a time where the upstream lab is not available. Floated the idea of pushing this back through the management chain and that did not go well. Unfortunately things did not work out and there will be some downtime. Hopefully the downtime will last no more than a month. Will likely go to Poughkeepsie to to the wiring. While the physical network infrastructure is there and ready, the main networking person there is on medical leave and has not been backfilled yet. There is also Security Stuff to deal with. There are a number of hurdles. It may be feasible to run some Jenkins builders in the cloud. Release order documentation =========================== From Joseph Mundackal (ref: https://ceph-storage.slack.com/archives/C05R975114L/p1763681398991599) Since we were doing a reef point release even though it's EoL, got asked what our criteria for deciding what release to do next were. Do we have that documented anywhere? According to Ilya, the reef release was a result of our n − 2 policy and tentacle being delayed. Expecting the delay, we did the work to get the PRs together and put a Reef out. The listed EoL date is when we expect the next major release to come out. Joseph asked if there would be interest in a PR to resolve the confusion between listed EoL date and actual release activity. Ilya suggested removing the column for expected EoL and replacing it with an explanatory note and agrees the current situation is probably confusing to users. Joseph will raise a PR to update the documentation. Reef Release ============ Yuri did a set of runs for the Reef release and sent an email to the leads asking to approve Erroneous Spam Marking ====================== Laura notes that when you find one of the emails in your spam filter you can mark it as Not Spam. Greg says you need to set up a folder and tell it never to send emails to that list to Spam. David Galloway mentions a few possible tweaks for deliverability next year.