Re: Out of FD - answer is not in FAQ

Hi!

I have a Fd problems too :

In my log file :
1999/09/17 08:15:25| sslReadClient: FD 98: read failure: (104) Connection reset
by peer
1999/09/17 08:15:25| sslReadClient: FD 203: read failure: (104) Connection
reset by peer
1999/09/17 08:15:25| sslReadClient: FD 31: read failure: (104) Connection reset
by peer
1999/09/17 08:15:25| sslReadClient: FD 44: read failure: (104) Connection reset
by peer
1999/09/17 08:15:25| sslReadClient: FD 57: read failure: (104) Connection reset
by peer
...
1999/09/17 08:17:20| comm_accept: FD 51: (105) No buffer space available
1999/09/17 08:17:20| httpAccept: FD 51: accept failure: (105) No buffer space
available
...
1999/09/17 08:23:38| Reserved FD adjusted from 100 to 832 due to failures
1999/09/17 08:23:51| Reserved FD adjusted from 861 to 882 due to failures
1999/09/17 08:23:55| WARNING! Your cache is running out of filedescriptors
1999/09/17 08:24:11| WARNING! Your cache is running out of filedescriptors
1999/09/17 08:24:27| WARNING! Your cache is running out of filedescriptors
...
If I'm run the cachemgr.cgi, then this say : in general runtime information :

File descriptor usage for squid:
        Maximum number of file descriptors: 1024
        Largest file desc currently in use: 326
        Number of file desc currently in use: 143
        Files queued for open: 0
        Available number of file descriptors: 881
        Reserved number of file descriptors: 882

My questions are :
1. What is the '...connection reset by perr' message ?
2. ...No buffer space available ??? This is a kernel bug, or squid bug ?
3. what is the 'reserved 882 file descriptor' ??? Why this happened ?

I'm use Linux 2.2.12 & memory-leak patch + Squid 2.2.stable4

Thanks in advance!

ps: the squid is very use the cpu...

Bye!Lajos
Received on Fri Sep 17 1999 - 06:29:51 MDT