Thanks Amos, my swap is 32GB, so that's causing the error as you said.
Which is the better choice: increase the swap size or reduce the cache_mem???2015-09-04 13:55 GMT-03:00 Amos Jeffries <squid3@xxxxxxxxxxxxx>:
On 4/09/2015 11:32 p.m., Jorgeley Junior wrote:
> Hi guys, I suspect my squid stop to serve request after rotate, in the
> morning, after I restarted it, everything goes to normal.
> here is the log:
> 2015/09/04 00:00:01 kid1| storeDirWriteCleanLogs: Starting...
> 2015/09/04 00:00:01 kid1| Finished. Wrote 39639 entries.
> 2015/09/04 00:00:01 kid1| Took 0.01 seconds (5804510.18 entries/sec).
> 2015/09/04 00:00:01 kid1| logfileRotate: stdio:/var/logs/store.log
> 2015/09/04 00:00:01 kid1| Rotate log file stdio:/var/logs/store.log
> 2015/09/04 00:00:01 kid1| logfileRotate: stdio:/var/logs/access.log
> 2015/09/04 00:00:01 kid1| Rotate log file stdio:/var/logs/access.log
> 2015/09/04 00:00:01 kid1| helperOpenServers: Starting 1/10
> 'basic_ncsa_auth' processes
> 2015/09/04 00:00:01 kid1| ipcCreate: fork: (12) Cannot allocate memory
As you can see Squid uses fork() to spawn its helpers. That means Linux
is going to allocate an N amount of virtual memory equal to the memory
currently being used by Squid.
> 2015/09/04 00:00:01 kid1| WARNING: Cannot run
> '/etc/squid-3.5.6/libexec/basic_ncsa_auth' process.
> 2015/09/04 00:00:55 kid1| WARNING: Memory usage at 67121 MB
> ...
Which is over 64 GiB.
Does your machine have 67,121 MB of virtual memory free ?
it would seem not to.
The only workaround for this is to keep Squid cache_mem small enough
that these oprations do not fail when it is fully in-use.
Amos
_______________________________________________
squid-users mailing list
squid-users@xxxxxxxxxxxxxxxxxxxxx
http://lists.squid-cache.org/listinfo/squid-users
--
_______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
