I'm looking at the `pdc_kth` config. <@UGJEZTL4D> ...
# configs
d
I'm looking at the
pdc_kth
config. @pontus work, thanks! I'm a bit puzzled by the plethora of partitions PDC provides and how that's taken care of by the config. First of all, the
shared
partition seems to be up to seven days, just like the
long
, so line 43 of https://github.com/nf-core/configs/blob/master/conf/pdc_kth.config could/should have a non-strict less than for number of days -- (and maybe also number of cpus) or? (I'm happy to make a PR for that, but am unsure if I'm interpreting things correctly.)
p
I'm not in on those discussions, and it's a bit of hear-and-guess but there is said to be ongoing discussions and the partition setup has definitely changed from when I first made it, so things can be wrong whenever. If that happens, make a PR. But here, since it breaks down anyway, line 43 could skip ``time < 7.d &&` since it doesn't matter, but I don't see that it hurts either (I don't recall if long allowed for longer jobs at the time, but now it seems to be the same).
d
Since it was a strict less than, my job that I specified for seven days ended up in the longer (I think)
long
queue. This is how I noticed this.
I'll make a PR for that.
Another note here, that I missed before, is that one would want to avoid the
long
partition as much as possible since it only allocates full nodes. Another reason for that PR, I think.