Menu
▾
▴
#449 Privoxy 3.0.2 leaks threads on RedHat 7.3
After having Privoxy always die on me I tried version 3.0.2.
It also dies, but in a very specific way: After a while I have 1024 Privoxy threads, causing all further connections to fail.
It is compiled with Pthreads.
Operating system RedHat Linux 7.3, GLibc 2.2.5-44.
Mozilla 1.4
I use the default config, with a few additional host entries.
Logged In: YES
user_id=322640
Are you chaining with another proxy by any chance?
What kernel version?
Is that our package or did you rebuild your own from sources?
I do not see either behavior here on RH7.2. Running since
Sept 22, with 5 threads right now. More info would be
helpful to narrow it down.
Logged In: YES
user_id=1205
Yes, it is running behind Squid 2.4.
The kernel is a vanilla 2.4.21.
The problem occurs both with the rpm and when I built from
source.
Logged In: YES
user_id=78811
Hal: Why did you ask for proxy chain? Known issue?
I've never seen this problem..
Logged In: YES
user_id=322640
Andreas, I am starting to suspect that all the reports of
Privoxy dying on RH logrotation are related to chaining
somehow. I cannot duplicate it here, and have Privoxy
rotating daily since last summer sometime (RH7.2). I am not
chaining though. The last several reports of that, have all
had been chained to Squid.
Logged In: YES
user_id=78811
Hal, that's interesting! I'll play with chaining and see if
I can reproduce either problem.
Logged In: YES
user_id=920479
I've been having exactly the same problems chaining with
Squid on Redhat 7.1 since Privoxy 3.0. I'm chaining in the
following order: Browser -> Squid 2.4 (transparent) ->
Privoxy 3.0.2. I have worked around the two problems by
setting logrotate to restart the daemon instead of reload
it, but this obviously doesn't fix the actual problems.
Logged In: YES
user_id=249980
Sorry for the (really, really, really) late response. We're
going through old bug reports and checking on their continued
validity. So... is this still an issue? Any workarounds we need
to know about? <Want to help?>
Logged In: YES
user_id=1205
I'm not using Squid anymore, but I still need to restart
Privoxy rather often as it refuses to do anything. I'm
currently using Privoxy 3.0.3 on Fedora Core 3.
Logged In: YES
user_id=249980
The next time it becomes unresponsive, check if it's hogging
100% of the CPU.
Logged In: YES
user_id=1312539
This Tracker item was closed automatically by the system. It was
previously set to a Pending status, and the original submitter
did not respond within 14 days (the time period specified by
the administrator of this Tracker).
I have the same problem, after having 1024 active connections to privoxy it start to use 100% of the cpu (only on one core). Privoxy is chained to TOR and it's using IPv4. All connections came from localhost.
Kernel 2.6.28.8-default1 #1 SMP x86_64
Privoxy version 3.0.11
Distribution: Debian lenny
It's not a low RAM memory problem, I'm running with 3.2GB + 5GB of swap.
Thanks for the report.
Could you rebuild Privoxy with debug symbols and attach
gdb once the problem appears, to see what's going on?
Does the problem eventually go away once the number
of active connections are reduced again?
Does Privoxy log any problems after you increased the debug level?
How are the 1024 active connections created?
This Tracker item was closed automatically by the system. It was
previously set to a Pending status, and the original submitter
did not respond within 14 days (the time period specified by
the administrator of this Tracker).