You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
At some point during runtime I suddenly didn't get any metrics for some processes anymore.
After adding some more debug output (see #79) I got this:
2019/01/15 13:10:05 error getting proc ID for pid 2434: integer overflow on token 2264576000
It seems this is because ProcStat in procfs/proc_stat.go uses too small integers for some fields.
in my case it is the vsize field that has the overflow.
The vsize should be unsigned as well.
Note that this happend on armv6, so 32bit.
The text was updated successfully, but these errors were encountered:
I've merged in the upstream changes, thanks! I've only done smoke testing on amd64 so far, but I expect you'll be pleased to hear that I'm now using pi zeros in my home cluster, and they're running process-exporter! So you should no longer be the sole finder of armv6/32-bit issues.
At some point during runtime I suddenly didn't get any metrics for some processes anymore.
After adding some more debug output (see #79) I got this:
It seems this is because
ProcStat
inprocfs/proc_stat.go
uses too small integers for some fields.in my case it is the
vsize
field that has the overflow.The
vsize
should be unsigned as well.Note that this happend on armv6, so 32bit.
The text was updated successfully, but these errors were encountered: