[SQU] FATAL: Too many queued redirector requests

I am using squid2.3stable1 (with squidGuard) in redhat6.2, kernel
2.2.14. we did not have any problems until one day i saw running
out of file descriptor messages in the cache.log. I did server
restarts; cleared the cache directory (if i did something wrong,
please point it out). but still the same problem.

so I followed Henrik's instruction on increasing the file descriptors.
now i have 4096 file descriptors available versus the original 1024.
(but I did not do a "echo 102432768 >
/proc/sys/net/ipv4/ip_local_port_range" since it was giving me the
following errors:
helperOpenServers: Starting 12 'squidGuard' processes
2001/02/23 13:50:05| commBind: Cannot bind socket FD 6 to 127.0.0.1:0: (98) Address already in use
2001/02/23 13:50:05| commBind: Cannot bind socket FD 6 to 127.0.0.1:0: (98) Address already in use
2001/02/23 13:50:05| ipcCreate: Failed to create child FD.
)

Anyway, squid would restart ever so often because of the following:
2001/02/26 09:24:32| WARNING: All redirector processes are busy.
2001/02/26 09:24:32| WARNING: 114 pending requests queued
FATAL: Too many queued redirector requests
Squid Cache (Version 2.3.STABLE1): Terminated abnormally.

Any suggestion on what i can do? Thanks.

--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html