pa...@arcor.de wrote:
----- Original Nachricht ----
Von: "Zeller, Jan" <jan.zel...@id.unibe.ch>
An: Amos Jeffries <squ...@treenet.co.nz>, Jeff Pang <pa...@arcor.de>
Datum: 04.08.2009 10:56
Betreff: AW: Aw: [squid-users] Running two squid3 process, Why?
Hi all,
would like to know if from user perspective the signal 6 will have a
significant http service interruption.
If I remember that correctly, kill SIGTERM and SIGINT will tell squid processes to exit gracefully.
Certainly kill SIGKILL will kill them forcely.
When receiving SIGTERM and SIGINT, the parent will kill a TERM to its child,
wait for child to exit (surely it's the time after child has finished the
current request handling), then parent exit gracefully.
So most time you just need to say kill `cat squid.pid` and squid will behave correctly.
This is what "squid -k shutdown" does?
use kill -9 `cat squid.pid` or "squid -k kill" to kill them forcely.
See also man kill under un*x.
Correct. "signal 6" is the Squid child process dying an unnatural
death due to some internal problem. Detected by a state assertion.
Those entries are from syslog correct? look in cache.log for the
matching timestamp records. You should see an assertion error message
with some more details about what has gone wrong which you can use to
look for a version where the problem is solved or mention in the bug
report to help get a fix made.
Amos
--
Please be using
Current Stable Squid 2.7.STABLE6 or 3.0.STABLE17
Current Beta Squid 3.1.0.12
