art-daq / artdaq_daqinterface

Other
0 stars 1 forks source link

DAQInterface kills processes across partitions on terminate #37

Closed eflumerf closed 2 years ago

eflumerf commented 2 years ago

This issue has been migrated from https://cdcvs.fnal.gov/redmine/issues/23015 (FNAL account required) Originally created by @jcfreeman2 on 2019-07-30 19:22:35


Just discovered on the ProtoDUNE cluster: when running in direct process mode, when DAQInterface is sent the terminate transition it can kill not only the artdaq processes associated with its partition, but also artdaq processes on OTHER partitions. This destructive interference needs to be fixed.

eflumerf commented 2 years ago

Comment by @jcfreeman2 on 2019-07-31 18:04:21


Resolved with commit 340b9089999ac639a17f2765f00d95901aa98051 at the head of feature/issue23015_kill_only_within_partition. Basically, to fix this all I needed to do was add the partition number into the grep performed when searching for artdaq processes to kill.

eflumerf commented 2 years ago

Comment by @jcfreeman2 on 2019-08-01 17:53:43


Reviewed by Kurt, who saw that when performing two runs simultaneously on partitions 4 and 5 on mu2edaq12 that terminating the partition 4 run had no effect on the partition 5 processes. I've merged the feature branch into the develop branch.