[squid-users] Re: tcp timeout issue

Also,

Here are the events leading up to the failures, not too sure if this
could of triggered it:

2007/10/01 22:40:39| parseHttpRequest: Unsupported method '1y¦íiÇNxRä-4gBx wZS
 k¨]5?À4&¸1y¼_éUXÄYfTh»ÒÎaäÇ4ÞáÓåÔÀ+Ü*
                                      q°`RCf«Ä­³ÇÂÝÚóVÃGy!×=
ÜÕ×Ôw)}
¢YUGªT9GET'
2007/10/01 22:40:39| clientReadRequest: FD 35 (70.231.40.8:1751) Invalid Request
2007/10/01 23:21:14| parseHttpRequest: CONNECT not valid in accelerator mode
2007/10/01 23:21:14| clientReadRequest: FD 19 (62.47.159.103:1814) Invalid Reque
st
2007/10/01 23:21:16| parseHttpRequest: CONNECT not valid in accelerator mode
2007/10/01 23:21:16| clientReadRequest: FD 39 (62.180.224.67:41740) Invalid Requ
est
2007/10/01 23:22:31| parseHttpRequest: Unsupported method 'TRACK'
2007/10/01 23:22:31| clientReadRequest: FD 18 (62.153.251.221:55305) Invalid Req
uest
2007/10/01 23:22:32| parseHttpRequest: Unsupported method 'TRACK'
2007/10/01 23:22:32| clientReadRequest: FD 33 (67.174.50.122:1676) Invalid Reque
st
2007/10/01 23:22:34| parseHttpRequest: Unsupported method 'CFYZ'
2007/10/01 23:22:34| clientReadRequest: FD 27 (89.55.123.193:63949) Invalid Requ
est
2007/10/01 23:24:17| parseHttpRequest: Unsupported method 'BADMTHD'
2007/10/01 23:24:17| clientReadRequest: FD 20 (89.52.178.46:64116) Invalid Reque
st
2007/10/01 23:24:18| parseHttpRequest: CONNECT not valid in accelerator mode
2007/10/01 23:24:18| clientReadRequest: FD 17 (82.228.47.219:2004) Invalid Reque
st
2007/10/01 23:24:18| parseHttpRequest: CONNECT not valid in accelerator mode
2007/10/01 23:24:18| clientReadRequest: FD 16 (85.183.133.10:59029) Invalid Requ
est
2007/10/01 23:24:18| parseHttpRequest: CONNECT not valid in accelerator mode
2007/10/01 23:24:18| clientReadRequest: FD 34 (82.98.89.2:63530) Invalid Request
2007/10/01 23:24:19| parseHttpRequest: CONNECT not valid in accelerator mode
2007/10/01 23:24:19| clientReadRequest: FD 40 (217.166.134.202:5735) Invalid Req
uest
2007/10/01 23:24:35| clientReadRequest: FD 47 (217.5.231.249:36369) Invalid Requ
est

On 10/2/07, Frank Ruiz <frank.ruiz@gmail.com> wrote:
> Greetings,
>
> I patched squid2.6 stable 14 with the tcp probe patch.
>
> It patched two files:
>
> cache_cf.c
> neighbors.c
>
> However, After about 14 hours of good runtime, my response times,
> began to suck, and began to see errors again indicative of the tcp
> probe issue:
>
> 2007/10/02 01:57:15| Detected REVIVED Parent: 10.10.10.20
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
> 2007/10/02 01:57:16| Detected DEAD Parent: 10.10.10.20
> 2007/10/02 01:57:16| TCP connection to 10.10.10.20/80 failed
>
>
> The origin server is available, however I keep getting
> revivied/connectionfailed/dead
>
> It seems that the only way to recover from this is a restart.
>
> I am running solaris 10, and I had to download the gnu patch utility
> in order to patch the src.
>
> Here was the patch applied.
>
> Index: src/cache_cf.c
> ===================================================================
> RCS file: /cvsroot/squid/squid/src/cache_cf.c,v
> retrieving revision 1.470
> diff -u -p -r1.470 cache_cf.c
> --- src/cache_cf.c 20 Jul 2007 21:08:47 -0000 1.470
> +++ src/cache_cf.c 28 Aug 2007 23:46:47 -0000
> @@ -1621,6 +1621,7 @@ parse_peer(peer ** head)
> p->stats.logged_state = PEER_ALIVE;
> p->monitor.state = PEER_ALIVE;
> p->monitor.interval = 300;
> + p->tcp_up = PEER_TCP_MAGIC_COUNT;
> if ((token = strtok(NULL, w_space)) == NULL)
> self_destruct();
> p->host = xstrdup(token);
> Index: src/neighbors.c
> ===================================================================
> RCS file: /cvsroot/squid/squid/src/neighbors.c,v
> retrieving revision 1.318
> diff -u -p -r1.318 neighbors.c
> --- src/neighbors.c 20 Jul 2007 21:08:47 -0000 1.318
> +++ src/neighbors.c 28 Aug 2007 23:46:47 -0000
> @@ -1010,12 +1010,13 @@ peerDNSConfigure(const ipcache_addrs * i
> debug(0, 0) ("WARNING: No IP address found for '%s'!\n", p->host);
> return;
> }
> - p->tcp_up = PEER_TCP_MAGIC_COUNT;
> for (j = 0; j < (int) ia->count && j < PEER_MAX_ADDRESSES; j++) {
> p->addresses[j] = ia->in_addrs[j];
> debug(15, 2) ("--> IP address #%d: %s\n", j, inet_ntoa(p->addresses[j]))
> ;
> p->n_addresses++;
> }
> + if (!p->tcp_up)
> + peerProbeConnect((peer *) p);
> ap = &p->in_addr;
> memset(ap, '\0', sizeof(struct sockaddr_in));
> ap->sin_family = AF_INET;
>
> Any ideas is much appreciated. Any special debug info you need, please
> let me know.
>
> Also, as I side note, I have monitorurl set as well
>
> cache_peer 10.10.10.20 parent 80 0 no-query no-digest originserver
> monitorinterval=30 monitorurl=http://10.10.10.20/test.jpg
>
> Thank you!
>
Received on Tue Oct 02 2007 - 12:46:02 MDT