pja237 / kp_oom Goto Github PK

Hi @pja237, I wanted to follow-up your post from apptainer/singularity#5850 (comment) with a few question. I figured an issue here would be more appropriate to continue.

1. Does this need to be rebuilt and reinstalled after every kernel upgrade?
2. What branch do you recommend to install from (I saw there was quite a few).
3. When this initiates killing a job how is it reported in Slurm? Does it show up as failed due to OOM or does it just show as canceled?
4. Do you have this deployed along side Open Science Grid (OSG)? If so has there been any issues?
5. Do you have any recommendations for monitoring when this has failed to kill a job? In our case this issue causes massive system performance degradation until eventually the system completely hangs. Our monitoring can already determine when the system hangs but we haven't figured how to detect this early before it gets to that point. I haven't seen a full crash and reboot yet.

Thanks you for making this public and taking the time to answer questions!