After connecting to Squid 1.1.5 via cachemgr.cgi (also 1.1.5) Squid
entered a state where our monitor shows 99.9 cpu utilization in *user*
mode. The access.log and cache.log show the cachemgr request (info) but
Squid doesn't log any further requests. Running squid -k shutdown results
in a two lines in cache.log announcing a shutdown and a 30 second wait for
connections to shut down. These log entries had the same time stamp as the
cachemgr request indicating that somehow the logging has not been updated
correctly. Squid then did not shut down connections within 30 seconds (I
waited for two minutes). I had to kill -KILL the squid process. After I
restarted squid everything worked fine (including the cachemgr interface).
I'm not reporting this problem in squid-bugs because so far I haven't been
able to reproduce it. However, three slight variations of this problem
occured within the last three days with version 1.1.2, 1.novm.4 and now
1.1.5. These occurences varied in the number of requests that were logged
after the cachemgr request and before Squid froze.
Any ideas? I would be especially interested in debug level suggestions.
Carlos
Received on Thu Jan 23 1997 - 21:01:05 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:34:10 MST