[09:23:19] greetings [09:54:57] morning [14:06:42] the more I look at the dumps distribution puppetization the more questions I have [14:17:13] easy to believe [14:17:37] my understanding is that there are next-gen dumps in the works (or already working? not sure) [14:17:48] not sure if/how that affects distribution [14:19:14] nope! [14:21:47] heh, I'll add a note to the agenda for tomorrow to briefly talk about dumps since it was on my mind too [14:22:58] taavi: I haven't messed with it much but I suspect it was intended to be a redundant system with two mounts and instead what we got was 2x as many chances for failure. [14:23:33] I don't think you'll get any pushback if you want to refactor things. [14:24:11] According to Riccardo the tools that use dumps can be counted on one hand; if that's true we should consider not supporting dumps at all on toolforge IMO. [14:26:01] fwiw the code to mount dumps on toolforge is basically the exact same as what mounts them on the rest of cloud vps, and the latest request to mount dumps on a new cloud-vps project is from this week so there are users relying on them [14:32:31] hmm, I just ran https://phabricator.wikimedia.org/P87831 on a k8s pod but I don't see that recorded as dumps access on the dashboard so the tracer might be a bit broken? [14:38:03] dhinus: do you think the triage meeting is necessary with the backlog being so tiny? [14:45:17] taavi: that sounds like a bug for Riccardo, want to open one? [14:45:56] I will [14:50:41] taavi: no we can skip it I think, I'm triaging the few ones we have right now [14:52:55] T415199 [14:52:57] T415199: Toolforge NFS tracing misses some dumps events - https://phabricator.wikimedia.org/T415199