nine sena_kun: the offer to host those bots on my server still stands. We could give them their own VM, so it'd be easy to duplicate their current setup without any major changes 10:00
sena_kun nine, we have plenty of resources (there is the blin server from our infra folks, your server, I can rent one without issues), just nobody to do the migration / setup work. 10:02
nine Why is docker necessary? 10:03
sena_kun I believe it makes maintaining easier (replicas, restarts), is more secure (that's when you execute someone's code on your machine), plus fixes some glibc issue if old enough distro is used. 10:06
nine systemd takes care of the first one and the other two are covered by having your own VM 10:10
sena_kun So be it, whatever floats the bot! 10:11
nine At least tkll we find someone with the time to properly care for them
MasterDuke nine: how much free ssd space does it have? 10:36
nine well, no ssd 10:38
MasterDuke hm, not even sure we've tested them on non-ssd. wonder if that would make dwarfs even more attractive since it would use the kernel's file cache, which would theoretically be more helpful for spinning rust 10:47
nine How much IO do these bots do? 11:01
rba - The blin server only runs docker container for security reason. 11:09
- I can take the lead with migration, but I need help.
- As I understand some bots are cpu hungry and some are disk hungry.
nine help in what form? 11:17
MasterDuke nine: not much sustained io. but lots of random access to extract the individual commit builds from their archives 11:46
rba: i wouldn't say any particular bot is cpu or disk hungry. it's all the builds that the bots use that take up a lot of disk space 11:47
of course depending on what code people run, a bot may use a bunch of cpu for a little while
timotimo docker isn't a security mechanism 13:25
moritz that's halfway true 13:36
docker makes it easy to use cgroups, which in turn can be a security feature
and namespaces for lots of stuff, like PIDs 13:37
Geth nqp/getlex_fallback: d81725d3ac | (Stefan Seifert)++ | src/vm/moar/QAST/QASTOperationsMAST.nqp
Fix try_get_bind_scope not stopping on undefined outer

The check for type BlockInfo always succeeds, because BlockInfo is actually not yet declared at that point, so it evaluates to NQPMu from which all types are derived. So the loop doesn't actually stop when outer is empty. Fix by checking for definedness instead.		 16:04
nqp/getlex_fallback: 64f05f8b8e | (Stefan Seifert)++ | 13 files
Support HLL fallback versions of ops looking for lexicals by name

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!