[08:56:50] <marostegui>	 s6 codfw hosts are now running bullseye
[09:08:54] <sobanski>	 \o/
[13:37:54] <Amir1>	 marostegui: FYI  I'm upgrading db1100 (s5) using cookbooks
[13:38:03] <marostegui>	 mmmm
[13:38:09] <marostegui>	 let me check if the schema change has run there
[13:38:15] <marostegui>	 ah, s5
[13:38:18] <marostegui>	 yeah, that's done
[13:38:19] <marostegui>	 I am doing s2
[13:38:20] <marostegui>	 so go for it
[13:38:23] <Amir1>	 yeah
[14:18:05] <Emperor>	 I appreciate "why" questions are dangerous, but... Why do we build our mariadb packages by taking an upstream binary tarball and then constructing a .deb from that (with a bunch of local changes - operations/software/dbtools/mariadb-client-deb) rather than by building from source?
[14:18:51] <jynus>	 we build it from source, or at least I did
[14:19:01] <volans>	 Emperor: "why" questions are always the best and you can challenge the status quo only in your first 6 months, after that you're part of the problem ;)
[14:19:25] <volans>	 so feel free to ask those too :)
[14:20:20] <Emperor>	 jynus: it is possible I have misunderstood how our packages are built, of course :) But that was the impression my spies had given me of how maros.tegui builds them...
[14:20:45] <Emperor>	 (and would explain why we have a script in dbtools to do just hat :)
[14:21:21] <jynus>	 I am pretty sure he did the same thing I did, there are references to that probably historic
[14:21:40] <jynus>	 from Sean or earlier
[14:23:11] <jynus>	 *that are
[14:23:55] <Emperor>	 I remember when I merged changes to mysql.service there was a problem with the next building not including the updated version.
[14:24:39] <jynus>	 sure, I remember that, I think he said he just hadn't rebased 
[14:25:04] <Emperor>	 Scope for automation, perhaps :)
[14:25:39] <jynus>	 there is always!
[14:26:13] <jynus>	 note that when I took over, I didn't have the resources to redo everything from zero
[14:26:35] <jynus>	 I was just happy with understanding how the previous person did it!
[14:27:31] <Emperor>	 Sure :) 
[14:28:24] <jynus>	 imagine that you enter, fillipo gives you the keys of swift and says goodbye, and you have no one else to ask about it
[14:28:48] <jynus>	 it was pure survival mode at that time
[14:29:04] <jynus>	 happily now there are more people that could do things better
[14:29:25] <Emperor>	 no, I am not going to imagine that, it would make me sad :)
[14:29:30] <jynus>	 in other words, "show me da patch" :-))))
[14:30:03] <Emperor>	 I should probably make sure I understand how we actually currently do it first...
[14:35:29] <jynus>	 If I remember correctly, I think a shortcut was taken to avoid storing the relatively large mysql source code on our repo, while still allowing for custom patches, not necesarilly in the cleanest way
[14:42:27] <marostegui>	 Emperor: I am in a meeting but TL;DR I talked to lukasz before you joined that I would love to hand over the package creation/automation to you as a debian dev :)
[14:42:48] <marostegui>	 thank you for volunteering!
[14:42:52] <Emperor>	 lolsob
[14:43:27] <jynus>	 you can take care of mine, too, I am more than happy to delegate that :-)
[14:43:36] <marostegui>	 and yes, I create them like the way jaime used to do it which is the way Sean used to do it!
[14:43:50] <jynus>	 it is turtles all the way down!
[14:44:12] <Emperor>	 I may have enough on my TODO right now :) But I should at some point understand exactly how we currently do it, and maybe I can propose some lower-friction approach
[14:44:26] <jynus>	 and Sean may not be even the original creator of that
[14:44:44] <jynus>	 I know he was of tendril, but he also inherited a lot of stuff
[14:45:55] <marostegui>	 let's all blame tim!
[14:47:15] <jynus>	 Emperor: I am semi-joking, I would be indeed happy to learn about DD stuff from you, but I don't want to be responsible for distracting you from more important things, eg. your OKRs, etc
[14:47:45] <jynus>	 maybe you could do a presentation or something?
[14:47:50] <jynus>	 at some point
[14:55:01] <Emperor>	 I think this might be a case where git-debrebase might be what we want; if that turns out to be the case, it might be worth a short talk on why it's useful , yes
[15:06:08] <Amir1>	 marostegui: I need to go eat lunch now but db1100 didn't get back up :(
[15:07:12] <marostegui>	 Amir1: I will check
[15:11:53] <marostegui>	 Amir1: it is now up, do you want me to start mysql or your script will take care of that?
[15:23:06] <Amir1>	 back
[15:23:15] <Amir1>	 marostegui: let me see if I can handle it
[15:23:20] <Amir1>	 what happened?
[15:25:40] <marostegui>	 no idea, the screen was blank and with weird symbols so I did a hard reset
[15:25:44] <marostegui>	 and it booted up just fine
[15:26:08] <Amir1>	 haha, okay
[15:26:32] <marostegui>	 I simply selected Linux instead of Windows on the bootloader and it was ok
[15:26:51] <Amir1>	 lol
[15:27:37] <Amir1>	 This didn't happen in codfw hosts, I'm gonna try another eqiad host to see if this is a one-time thing or the cookbook simply can't handle eqiad
[15:28:48] <marostegui>	 it is not weird that hosts have troubles booting up
[15:29:04] <marostegui>	 especially those which are old or have been on for a long time
[15:29:46] <Amir1>	 yeah...
[15:52:06] <Emperor>	 ...why was the bootloader even offering Windows?
[15:58:08] <sobanski>	 Windows ME
[16:06:59] <Emperor>	 😱
[19:57:57] <Amir1>	 jynus: the backup sources for several sections (s1, s4, s8) is not replicating both in codfw and eqiad. I didn't find anything in tickets nor SAL. Is it intentional?
[19:58:06] <Amir1>	 I hope I didn't break anything
[21:13:50] <marostegui>	 it is
[21:15:36] <marostegui>	 I talked to amir in private like an hour ago, but just mentioning it here so other readers get to know it is the way it works :)