[13:40:01] Emperor: apologies if you documented this somewhere and I missed it, but what do you see as the workflow for packaging software developed at the wmf? [13:40:30] should we building them as Debian native packages? [13:44:42] I'm conditioned to cringe whenever I see someone shipping a `debian/` directory, but maintaining a separate branch for packaging seems like unnecessary ceremony for something internal [13:49:39] depends which workflow you follow ;) [13:49:52] is that python or go? [13:50:07] gerrit or gitlab? [13:50:34] well... it's go, and gitlab [13:50:48] I worry about to cringe over something else... [13:51:06] then I think you could build the package directly on gitlab on the trusted workers [13:51:46] but I don't know if there is already some standardization for go repos over there [13:51:55] and yes debian package is the preferred way in general [13:52:09] for deploying to a host/vm ofc [13:52:31] yeah, I mean the way it is packaged [13:53:00] I agree that a Debian native package would seem to be the obvious approach [13:54:01] Emperor: ok, with regard to CI, would you then just merge main into your *-wikimedia branch when read to trigger a package build? [13:54:11] s/read/ready/ [13:54:31] that would work, yes, or if you wanted to always build the tip of it, rename the default branch. [13:54:39] https://wikitech.wikimedia.org/wiki/Debian_packaging/Tutorial#Executive_Summary_/_TL;DR [14:13:15] volans: sorry, I was preparing to cringe because I thought you were going to say that it was necessary that the packages be different based on source lang or build env 😃 [14:13:33] lol [16:36:17] Emperor: o/ me and Jesse are still working on some remaining bugs/features of UEFI, but the ms-be nodes should be hopefully ready soon-ish. In any case, for thanos-be if there is a big urgency we can just use what we have now [16:43:48] elukey: thanks for the update (and for all the work!) I think it'd be great if we could get the new thanos-be nodes ready ASAP; I'm happy to wait on the ms-be nodes to let you hammer out the remaining issues. [16:44:48] (does that make sense?) [16:48:49] Emperor: if they are already racked and ready for provision we could go ahead even tomorrow [16:56:02] That would be super exciting, although I'm not going to make ring changes until Monday now in all likelihood [16:58:02] sure sure, but the ring is not going to change with a reimage right? It is like for ms-be, the hosts are ready to go but they need manual action [16:58:17] if so, if thanos-be is in the DC, we can definitely work on those [16:58:20] lemme check netbox [16:59:17] Sorry, the ring change is the final step of bringing a new storage node into the cluster and applying load to it. (cf https://wikitech.wikimedia.org/wiki/Swift/How_To#Add_a_storage_node_to_the_cluster ) [16:59:17] ah ok so https://phabricator.wikimedia.org/T370452 seems not rakced yet