Message boards : News : killing extremely long SixTrack tasks
Message board moderation

To post messages, you must log in.

AuthorMessage
Alessio Mereghetti
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 29 Feb 16
Posts: 157
Credit: 2,659,975
RAC: 0
Message 39150 - Posted: 18 Jun 2019, 16:49:08 UTC

Dear all,

we had to kill ~10k WUs named:
w-c*_job*__s__62.31_60.32__*__7__*_sixvf_boinc*

due to a mismatch between the requested disk space and that actually necessary to the job.
These tasks would anyway be killed by the BOINC manager at a certain point with an EXIT_DISK_LIMIT_EXCEEDED message - please see:
https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=5062
for further info.

These tasks cover 10^7 LHC turns, a factor 10 larger than usual, with files building up in dimension until the limit is hit.

The killing does not involve all tasks with such names - I have killed only those that should cover the stable part of the beam; these tasks are expected to last long and hence reach the limit in disk usage. The other WUs should see enough beam losses that the limit is not reached - please post in this thread if this is not the case. The cherry-picking killing was done in the effort of preserving as much as possible tasks already being crunched or pending validation.

As soon as you update the LHC@project on your BOINC manager you should see the task being killed.

We will resubmit soon the same tasks, with appropriate disk requirements.
Apologies for the disturbance, and thanks for your understanding.
A.
ID: 39150 · Report as offensive     Reply Quote

Message boards : News : killing extremely long SixTrack tasks


©2024 CERN