<gchristensen>
it only marks it as started once it has a free evaluation slot, and it typically has three evaluation slots
<Mic92>
gchristensen: is it possible to do this faster/
<Mic92>
?
<Mic92>
As long as it shows pending the action terminates
<Mic92>
I can try to increase the timeout in the meantime
<gchristensen>
hmm
<gchristensen>
yes it could happen sooner, certainly. for example, the job to start the evaluation could be tee'd off in to another queue which just marks them as started
<gchristensen>
it is unfortunate evaluation requires such an extreme amount of memory -- otherwise we could have far more evaluators, and never have a delay
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos-borg
<Mic92>
We need to make nixpkgs evaluation a crypto currency
orivej has quit [Ping timeout: 260 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-borg
<hexa->
and get rich or die tryin'?
<gchristensen>
lol
<Mic92>
I hope 30 min is enough in most cases. There will be case where it won't be.
<Mic92>
But in the end it should not break anything. It should be clear that if the service timeouts but ofborg still runs through that the PR should not be blocked.
<Mic92>
gchristensen: is 32 gb enough for an evaluator?
<gchristensen>
I believe that is what we allocate now, yeah
<Mic92>
Are those 32 gb then exclusivly used? I have also one machine in university that is largely unused that has 16 gb.
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nixos-borg
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-borg
<LnL>
gchristensen: hmm, where do you get 32 from?