Hi!

After seeing that this problem clearly still exists on 6.12, I decided to
take some numbers, the machine originally had 4 GB of RAM but I upgraded it
to 8 in order for us to clearly see this:

# dpkg -l|grep squid
ii  squid                           6.12-1                         amd64        
Full featured Web Proxy cache (HTTP proxy GnuTL
ii  squid-common                    6.12-1                         all          
Full featured Web Proxy cache (HTTP proxy) - co
ii  squid-langpack                  20220130-1                     all          
Localized error pages for Squid
ii  squidclient                     6.12-1                         amd64        
Full featured Web Proxy cache (HTTP proxy) - HT
# uptime
 12:04:12 up 2 min,  1 user,  load average: 1,31, 0,60, 0,23
# free
               total        used        free      shared  buff/cache    
available
Mem:         8131288     1139860     6938544        6328      305880     6991428
Swap:        1047548           0     1047548
# ps aux|grep squid
root         813  0.0  0.2  60284 18992 ?        Ss   12:01   0:00 
/usr/sbin/squid --foreground -sYC
proxy        830 76.4  5.8 768416 471616 ?       R    12:01   2:13 (squid-1) 
--kid squid-1 --foreground -sYC
proxy        843  1.4  0.0   5644  1900 ?        S    12:01   0:02 
(logfile-daemon) /var/log/squid/access.log

That's two minutes after booting, but if we wait for a while... this is 21h
after booting:

# uptime
 08:58:23 up 20:57,  1 user,  load average: 1,41, 1,32, 1,22
# free
               total        used        free      shared  buff/cache    
available
Mem:         8131288     5452228      183520        6660     2798416     2679060
Swap:        1047548        1036     1046512
# ps aux|grep squid
root         813  0.0  0.2  60284 18872 ?        Ss   ene28   0:00 
/usr/sbin/squid --foreground -sYC
proxy        830 33.1 56.5 4688628 4600644 ?     R    ene28 416:24 (squid-1) 
--kid squid-1 --foreground -sYC
proxy        843  0.6  0.0   5644  1900 ?        S    ene28   8:42 
(logfile-daemon) /var/log/squid/access.log

You can clearly see how squid has grown, but I believe those 21 hours were
our low usage part of the day, then we went to the high usage part and...

# uptime
 10:34:00 up 22:32,  1 user,  load average: 1,38, 1,37, 1,23
# free
               total        used        free      shared  buff/cache    
available
Mem:         8131288     7809592      269576        1276      324884      321696
Swap:        1047548      954424       93124
# ps aux|grep squid
root         813  0.0  0.0  60284  7424 ?        Ss   ene28   0:00 
/usr/sbin/squid --foreground -sYC
proxy        830 37.5 69.5 6580768 5658884 ?     R    ene28 508:26 (squid-1) 
--kid squid-1 --foreground -sYC
proxy     707076  2.0  0.0   5644  1892 ?        S    09:34   1:12 
(logfile-daemon) /var/log/squid/access.log
root      740005  0.0  0.0   6352  2144 pts/0    S+   10:34   0:00 grep squid

# uptime
 11:45:02 up 23:43,  1 user,  load average: 1,08, 1,03, 1,01
# free
               total        used        free      shared  buff/cache    
available
Mem:         8131288     7739792      268192        2248      397008      391496
Swap:        1047548      896788      150760
# ps aux|grep squid
root         813  0.0  0.0  60284  7384 ?        Ss   ene28   0:00 
/usr/sbin/squid --foreground -sYC
proxy        830 40.2 77.3 7172836 6292852 ?     R    ene28 572:31 (squid-1) 
--kid squid-1 --foreground -sYC
proxy     707076  1.7  0.0   5644  1832 ?        S    09:34   2:16 
(logfile-daemon) /var/log/squid/access.log
root      778771  0.0  0.0   6352  2036 pts/0    S+   11:45   0:00 grep squid

# uptime
 13:34:41 up 1 day,  1:33,  1 user,  load average: 1,35, 1,08, 1,03
# free
               total        used        free      shared  buff/cache    
available
Mem:         8131288     7951444      207144        1712      223652      179844
Swap:        1047548     1041812        5736
# ps aux|grep squid
root         813  0.0  0.0  60284  5132 ?        Ss   ene28   0:00 
/usr/sbin/squid --foreground -sYC
proxy        830 43.7 86.7 8077792 7051160 ?     R    ene28 670:25 (squid-1) 
--kid squid-1 --foreground -sYC
proxy     707076  1.6  0.0   5644  1264 ?        S    09:34   3:58 
(logfile-daemon) /var/log/squid/access.log
root      839121  0.0  0.0   6352  2108 pts/0    S+   13:34   0:00 grep squid

We got to the oom-killer at 14:36:51

[95735.290642] [    813]    13   813    15071      499   110592     1834        
     0 squid
[95735.292468] [    830]    13   830  2095703  1842365 16805888   235203        
     0 squid
[95735.374533] 
oom-kill:constraint=CONSTRAINT_NONE,nodemask=(null),cpuset=/,mems_allowed=0,global_oom,task_memcg=/system.slice/squid.service,task=squid,pid=830,uid=13
[95735.378340] Out of memory: Killed process 830 (squid) total-vm:8382812kB, 
anon-rss:7368496kB, file-rss:696kB, shmem-rss:0kB, UID:13 pgtables:16412kB 
oom_score_adj:0

As you can see, we are running just with 1 worker, in the past this has
shown more stability. Config file is pretty simple, we are also just doing
connects for the ssl part, no decription or anything.

We've had this bug on Bookworm, but I'd really love to see this fixed for
trixie, so if you somebody wants to take a deeper look and needs more info,
wants special tests or whatever, feel free to contact me and I'll try to
setup what is needed.

Regards.
-- 
Manty/BestiaTester -> http://manty.net

Reply via email to