The fix is likely in v14.11.2 as you noted below.
Quoting Christopher B Coffey <[email protected]>:
Hi, Slurmctld suffered a segfault yesterday for the first time this year. Is this related to the new version 14.11.1 ? Nothing really was happening at the time except for ~200 jobs being submitted, hardly a lot. kernel: slurmctld_srvcn[15014]: segfault at 795 ip 0000000000438fb0 sp 00007fe6410cedb8 error 4 in slurmctld[400000+1cb000] Maybe this bug has been addressed? -- Fix issue with association hash not getting the correct index which could result in seg fault. Also, not sure if it’s related to 14.11.1 too, but while having a reservation in place for upcoming maintenance, some jobs that are deferred show this state: (Resources) instead of (ReqNodeNotAvail(Unavailable:)). I don’t believe we saw this happen in 14.03. Maybe this bug has been addressed? -- Improve information recording for jobs deferred due to advanced reservation. Thanks! — Christopher Coffey High-Performance Computing Northern Arizona University 928-523-1167
-- Morris "Moe" Jette CTO, SchedMD LLC Commercial Slurm Development and Support
