Something funny (peculiar) happened last week.
After running squid-1.2beta22 for a week or so (previously beta20) one of
the caches starting using 100% CPU (well, as near to it as possible).
No big deal, I just restarted it and assumed something evil had gone wrong.
I now find, coming in this morning, both active caches are using 100% CPU,
which is unusually high for caches which normally using at most 5% or so
when quite busy.
They are still functioning, and still fairly responsive, so I'll leave them
a little longer, but if anyone has any suggestions on what to check before I
restart them, please get in quick.
I'll probably grab an strace before restarting, and I notice I'm presently
doing about 6600 select loop/s, which indicates perhaps on of the FDs is
hosed and squid hasn't realised?
I'll rebuild with debugging symbols to next time this happens I can attach
with gdb and try to figure out what FD is screwing select up....
Suggestions or comments?
-cw
Received on Tue Jul 29 2003 - 13:15:51 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:11:49 MST