search

reanahub / reana-demo-atlas-recast

REANA example - RECAST analysis based on ATLAS software stack
MIT License
4 stars 33 forks source link

workflows: less strict memory limits #39

Closed tiborsimko closed 2 years ago

tiborsimko commented 2 years ago

Uses 256Mi instead of 64Mi which was not enough on some REANA installations such as the the REANA production server at CERN.

tiborsimko commented 2 years ago

Haven't looked in detail, but between local Kind installations and cloud K8s 1.19 or 1.21 installations and job controller with/without HTC/HPC support (albeiit this may not play a role), it is safer to go higer. E.g. when I tried 128Mi, most of jobs were passing OK, but I had a few that failed on some systems... So going up to 256Mi which is very safe, even though it may be a bit of an overkill.