> "kill -9" will definitely leave some things in an odd state. Try the
> double-kill first.
On Sun, 13 Mar 2022 18:38:05 -0400
Alex Rousskov <rousskov@xxxxxxxxxxxxxxxxxxxxxxx> wrote:
Just to add: kill -9 will not kill SMP Squid workers and other kid
processes. To a large degree, Squid will just keep running as if nothing
happened -- that signal cannot be caught and processed specially by the
receiving Squid process. Do not use "kill -9 `cat pidfile`" if you are
using SMP Squids!
On 13.03.22 18:49, Dave Blanchard wrote:
It turns out double kill with a SIGTERM, besides being rather inelegant,
doesn't do the job either. It takes its sweet ass time--a few seconds--to
exit.
this is how squid works - it needs to cleanly close connections etc.
you can lower shutdown_lifetime option in config file
One of the main purposes for Squid on my system is to proxy Chromium,
which likewise has the brilliant design of continuing to load page
elements (on my extremely low bandwidth connection with a quota) when I've
clearly told it to STOP NOW. The point is to put a hard, immediate stop
to it, via killing the proxy.
try different browser probably. I don't know if chromium has option not to
fetch images etc.
So is there a signal that can be sent which is the exact equivalent of
squid -k interrupt? The only reason I was using that command in the first
place is because 'kill' wasn't doing the job correctly.
either shut it down correctly so it can start again, or don't shut it down
and then expect not being able to start.
Why, oh why, does the config file have to be 100% valid, when the request
is to simply STOP the damn thing? That makes no logical sense.
this makes much sense and has already been explained.
starting with invalid configuration ignorec means possible unexpected
behaviour.
according to what you wrote above, I am sure you don't want to squid fetch
full content of interrupted transactions just because config directives were
not read properly.
Deciding
to crash on some obscure malformed config directive while simply giving
the command to STOP is NOT the right way to handle the situation. Forget
about "not enough resources to add that feature." That is how it should
have been designed from day one! So basically, "not enough resources to
fix what we designed wrong in the first place." To be fair, I guess you
just initially assumed everyone would want the proxy to finish up what
it's doing before quitting. But that's not true in every case.
What is actual use of the pidfile to the end user, when kill -9 does not
actually kill the damn thing,
kill -9 DOES kill the damn thing, and it doesn't allow the damn thing
cleaning up the mess.
https://porkmail.org/era/unix/award#uuk9letter
but actually fucks things up worse by
leaving zombie processes running in the background--and SIGTERM doesn't do
the job either, even when asked twice? Why the hell aren't all signals
being forwarded to the child process(es), or otherwise architected so
death of the controlling process brings down the whole thing?
this discussion makes no sense.
if you don't like how squid works, try e.g. wwwoffle.
--
Matus UHLAR - fantomas, uhlar@xxxxxxxxxxx ; http://www.fantomas.sk/
Warning: I wish NOT to receive e-mail advertising to this address.
Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu.
99 percent of lawyers give the rest a bad name.
_______________________________________________
squid-users mailing list
squid-users@xxxxxxxxxxxxxxxxxxxxx
http://lists.squid-cache.org/listinfo/squid-users
