Vivek wrote:
Amos,
I didn't see any errors related to squid restart/crash in cache.log. Is
there any way to debug this ?
Other than dumping the WCCP info display on the router I have no idea.
As Adrian said, its probably just a temporary issue with a long index
rebuild causing Squid to ignore one of the WCCP announcements.
If so the loss time will remain the same no matter how frequently or
infrequently you rotate. Thus less often will be less outage time.
-----Original Message-----
Tom Penndorf wrote:
Hello,
Vivek schrieb:
Amos,
I am rotating the logs using " squid -k rotate ". In the crontab,
0 */1 * * * /usr/local/squid/sbin/squid -k rotate
I think, this will cause squid to do an reload, so it will shortly
stop
the wccp-service. Perhaps, you should rotate the files only once a
day.
"-k rotate" was the right way to do it. Only reloads the helpers and
rotates the logs.
WCCP and other public-facing services should not be seeing any loss of
service at all on rotate. Worst case is a short[1] lag while the store
index journal gets rebuilt and dumped to disk.
[1: for various definitions of 'short'.]
Is there any sign in cache.log of Squid crashing at or around those times?
2.7 has an auto-restart built in that may be masking it from notice.
Tom
squid version - squd-2.7 Stable 6
IOS version - 12.4(15)T6
Amos
--
Please be using
Current Stable Squid 2.7.STABLE6 or 3.0.STABLE18
Current Beta Squid 3.1.0.13
