Closed linas closed 7 years ago
What caught my eye was the long string of Trace messages, immediately before the crash. These never normally get printed, so I assume they are crash-related.
FWIW, I am seeing strange and unexpected crashes/hangs on this machine; there's some chance that its hardware-related.
A memory corruption happened, and the question is whether it ior hardware fault,. These type of servers usually has ECC memory, so memory errors are less common than in desktop computers.
See if the kernel log reports hardware errors. In any case, I don't know what further can be done with this issue.
yeah. I thought I'd log this, but its junk
I might have fixed the root cause of this in #495. Other crashes remain (this machine has a very very old glibc on it, maybe 10 years old, and TLS seems to be unhappy.)
This is a very very rare error: I've seen it once, only, after about 8-12 cpu-hours of processing: