cole-h has quit [Ping timeout: 264 seconds]
tilpner has quit [Read error: Connection reset by peer]
tilpner has joined #nixos-infra
cole-h has joined #nixos-infra
tilpner_ has joined #nixos-infra
tilpner has quit [Ping timeout: 256 seconds]
tilpner_ is now known as tilpner
cole-h has quit [Ping timeout: 246 seconds]
henrietaMag has quit [Ping timeout: 260 seconds]
FRidh has joined #nixos-infra
<
gchristensen>
FRidh: I'm inclined to cancel all non-current queued builds to avoid building backlog
<
FRidh>
gchristensen: actually should have a separate ci group
<
gchristensen>
I wonder also if we should re-examine the scheduler "points"?
<
FRidh>
yes, I do that quite often
<
FRidh>
in this case its staging-20.09
<
FRidh>
I think we've not been very careful with the resources we have
<
FRidh>
pushing to that branch often, causing rebuilds
<
FRidh>
instead of grouping them in batches
<
gchristensen>
I agree
adisbladis has joined #nixos-infra
<
FRidh>
points is for sure an issue as well, but I don't think it helps here so much. release gets more, which is good
<
FRidh>
if we talk about points it is the crazy high amount for haskell that should be decreased
<
FRidh>
I mentioned also to mweinelt to use the "staging (stable)" project to group PR's for staging-20.09
<
lukegb>
If only we had BORS...
<
FRidh>
We need to document that better, or maybe limit somehow differently the pushing to that branch
<
FRidh>
but for aarch64, realistically, it cannot build a package set this big with the resources we have
<
gchristensen>
I will try to get us some more machines we can use
<
gchristensen>
I have several machines available to us, but nixos does not run on them today -- maybe we can find a volunteer to make that work
<
adisbladis>
gchristensen: What does run on them?
<
gchristensen>
I imagine the rescue-mode alpine works on them
<
gchristensen>
they're fairly exotic hardware and not, uh, wellk nown
<
adisbladis>
gchristensen: I'm thinking if we actually need to run NixOS on the metal
<
adisbladis>
Or if a container/chroot/whatever would work
<
gchristensen>
dunno :P want me to get you access to one?
<
adisbladis>
No guarantees of any results ;)
<
gchristensen>
I have like a dozen of them, I can hand them out :P
<
adisbladis>
But I think it's interesting to see what could be done with minimal effort
<
gchristensen>
but we can't afford half-assed stuff, the maintenance of half-assed is too high imo
<
gchristensen>
(not that the idea is inherently half-assed, but I wouldn't want to manually turn-up a bunch of machines we don't have tooling for)
<
FRidh>
if only nixbuild.net had aarch64
<
adisbladis>
gchristensen: Of course
<
FRidh>
anyway, I think I'll then set meta.schedulingPriority for the blocking job
<
FRidh>
aarch64 users of nixpkgs-unstable will then have to accept that most things just arent build
<
gchristensen>
are there any special features on that derivation?
<
FRidh>
not that I can tell
tilpner_ has joined #nixos-infra
tilpner has quit [Ping timeout: 256 seconds]
tilpner_ is now known as tilpner
cole-h has joined #nixos-infra
henrietaMag has joined #nixos-infra
FRidh has quit [Quit: Konversation terminated!]
cole-h has quit [Quit: Goodbye]
cole-h has joined #nixos-infra
lassulus has quit [Remote host closed the connection]
lassulus has joined #nixos-infra
zimbatm has quit [Ping timeout: 272 seconds]