lally has quit [Read error: Connection reset by peer]
lally has joined #haskell.nix
dhess has quit [*.net *.split]
NinjaTrappeur has quit [*.net *.split]
shapr has quit [*.net *.split]
ocharles has quit [Ping timeout: 264 seconds]
ocharles has joined #haskell.nix
dhess has joined #haskell.nix
shapr has joined #haskell.nix
NinjaTrappeur has joined #haskell.nix
stites[m] has quit [Ping timeout: 246 seconds]
Ericson2314 has quit [Ping timeout: 246 seconds]
michaelpj has quit [Ping timeout: 246 seconds]
domenkozar[m] has quit [Ping timeout: 260 seconds]
ptival[m] has quit [Ping timeout: 260 seconds]
jonge[m] has quit [Ping timeout: 260 seconds]
immae has quit [Ping timeout: 240 seconds]
siraben has quit [Ping timeout: 268 seconds]
alexarice[m] has quit [Ping timeout: 240 seconds]
domenkozar[m] has joined #haskell.nix
jonge[m] has joined #haskell.nix
michaelpj has joined #haskell.nix
Poscat[m] has joined #haskell.nix
siraben has joined #haskell.nix
immae has joined #haskell.nix
alexarice[m] has joined #haskell.nix
hekkaidekapus} has quit [Remote host closed the connection]
hekkaidekapus} has joined #haskell.nix
hekkaidekapus} has quit [Ping timeout: 268 seconds]
hekkaidekapus} has joined #haskell.nix
hekkaidekapus} has quit [Ping timeout: 268 seconds]
hekkaidekapus} has joined #haskell.nix
pjb has quit [Remote host closed the connection]
pjb has joined #haskell.nix
fendor has joined #haskell.nix
__monty__ has joined #haskell.nix
fendor_ has joined #haskell.nix
fendor has quit [Ping timeout: 256 seconds]
hekkaidekapus} has quit [Ping timeout: 268 seconds]
hekkaidekapus} has joined #haskell.nix
<dhess>
infinisil: I'm perplexed by that issue, especially this bit: "with the problem disappearing if the store path that's missing the files was rebuilt a couple times."
<dhess>
Any idea why that was happening? It seems like it should be possible. Was it a race condition in the build or something?
<dhess>
/should/shouldn't/
<dhess>
I'm still getting very strange errors related to that using yesterday's haskell.nix, but only on one machine (our Hydra). It makes me wonder if the Nix store on that machine is in a bad state.
<michaelpj>
seems like it must be something like that. Somehow the build sometimes produces the wrong output, or the output gest corrupted or something. It's very perplexing
<michaelpj>
yeah, your store can end up effectively corrupted
<dhess>
ugh, this is our Hydra so it really sucks.
<dhess>
I guess I could just do a massive GC and let it refill from substituters.
<dhess>
except that won't work if these packages are in our S3 binary cache.
<dhess>
but no, the builder where this is working is using our S3 binary cache and it's not having any problems, so maybe our cache is OK
NinjaTrappeur has quit [Quit: WeeChat 3.0]
NinjaTrappeur has joined #haskell.nix
hekkaidekapus} has quit [Remote host closed the connection]
fendor_ is now known as fendor
hekkaidekapus} has joined #haskell.nix
<michaelpj>
yeah it's a huge pain if you end up with wrong outputs in the cache :(
<michaelpj>
we had this with cachix before and I had to ask domen to do some cleanup for me, don't know how you do that in hydra
<dhess>
me neither. I think we're OK for now though. I suspect what is happening is that these broken packages are built during evaluation (that is where we see all the errors) and then their dependencies die in these unexpected ways, also during eval. So we don't get to the stage where anything broken is actually built and put in the binary cache.
<michaelpj>
yikes
<michaelpj>
well, we put in some guards to hopefully make it fail instead of producing bad output
hekkaidekapus} has quit [Ping timeout: 268 seconds]
hekkaidekapus} has joined #haskell.nix
<ocharles>
dhess: I've had a lot of luck using a post-build-hook to upload things to the cache. That happens during evaluation time, too
<dhess>
ocharles: are you all even using a Hydra? Or has your buildkite plugin replaced any need for it?
<ocharles>
We are almost entirely on Buildkite now, but spent years on Hydra
<ocharles>
I don't think anything important is on Hydra now
<dhess>
I am sort of moving in that direction now. Hydra is pretty good for something that's free, but I spend quite a bit of my "maintenance time" on it.
acarrico has joined #haskell.nix
acarrico has quit [Ping timeout: 264 seconds]
hekkaidekapus{ has joined #haskell.nix
hekkaidekapus} has quit [Ping timeout: 268 seconds]
acarrico has joined #haskell.nix
__monty__ has quit [Quit: leaving]
fendor has quit [Remote host closed the connection]