ip_conntrack locks up hard on 2.4.0 after about 10 hours

ip_conntrack locks up hard on 2.4.0 after about 10 hours

[safemode]

It seems that for one reason or another, ip_conntrack totally locks (not
removeable) after about 10 hours of continued use.  All i found were
these messages in my dmesg output
Jan  6 06:18:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
when=0x5d9e, caller=c01a6bf1
Jan  6 06:18:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
when=0x5b2f, caller=c01a6bf1
Jan  6 06:18:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
when=0x56bb, caller=c01a6bf1
Jan  6 06:40:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
when=0x217db, caller=c01a6bf1
Jan  6 06:40:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
when=0x2363e, caller=c01a6bf1
Jan  6 06:40:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
when=0x21b64, caller=c01a6bf1
Jan  6 06:40:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
when=0x1fa85, caller=c01a6bf1

This makes it impossible to make any sort of network socket connection
and all prior connections died.  As i said you cannot remove the module
to reset ip_conntrack and i'm not sure what could have caused this as it
did work up until i woke up this morning, with a total running time of
about 10 hours or so.  I'd consider this bug rather important, if anyone
thinks this is not an ip_conntrack bug and rather something that has
changed that i havn't read about, help would be nice. :)    i have been
using iptables since it came out though and ip_conntrack has only been
bad once before,   on test5 when it wouldn't kill old dead socket
connections and eventually starved itself of free sockets.



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: ip_conntrack locks up hard on 2.4.0 after about 10 hours

[safemode]

[-- Attachment #1: Type: text/plain, Size: 2059 bytes --]


Setiathome 3.03 and 3.x most likely causes the ip_conntrack errors which
quickly brings the system to a screetching network halt.   I suggest nobody
run setiathome on their firewall/gateway/router if they're using iptables
with 2.4.x.   Not sure how it causes this error nor would it matter to me
since i wouldn't be able to recode the client anyway.  I'm sure there are
setiathome developers (at least one) paying attention to this list.  The
client is broken.




safemode wrote:

> It seems that for one reason or another, ip_conntrack totally locks (not
> removeable) after about 10 hours of continued use.  All i found were
> these messages in my dmesg output
> Jan  6 06:18:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
> when=0x5d9e, caller=c01a6bf1
> Jan  6 06:18:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
> when=0x5b2f, caller=c01a6bf1
> Jan  6 06:18:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
> when=0x56bb, caller=c01a6bf1
> Jan  6 06:40:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
> when=0x217db, caller=c01a6bf1
> Jan  6 06:40:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
> when=0x2363e, caller=c01a6bf1
> Jan  6 06:40:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
> when=0x21b64, caller=c01a6bf1
> Jan  6 06:40:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
> when=0x1fa85, caller=c01a6bf1
>
> This makes it impossible to make any sort of network socket connection
> and all prior connections died.  As i said you cannot remove the module
> to reset ip_conntrack and i'm not sure what could have caused this as it
> did work up until i woke up this morning, with a total running time of
> about 10 hours or so.  I'd consider this bug rather important, if anyone
> thinks this is not an ip_conntrack bug and rather something that has
> changed that i havn't read about, help would be nice. :)    i have been
> using iptables since it came out though and ip_conntrack has only been
> bad once before,   on test5 when it wouldn't kill old dead socket
> connections and eventually starved itself of free sockets.
>




[-- Attachment #2: Type: text/html, Size: 2432 bytes --]

Re: ip_conntrack locks up hard on 2.4.0 after about 10 hours

[David S. Miller]

   Date: 	Sat, 06 Jan 2001 10:37:54 -0500
   From: safemode <safemode@voicenet.com>

   Jan  6 06:18:10 icebox kernel: reset_xmit_timer sk=c17fd040 1
   when=0x5d9e, caller=c01a6bf1

I posted a fix for this on Linux-kernel yesterday, had you tested it
you would have seen at least this part of your problem report go away.
I'm reposting the fix for your convenience:

--- net/ipv4/tcp_input.c.~1~	Wed Dec 13 10:31:48 2000
+++ net/ipv4/tcp_input.c	Fri Jan  5 17:01:53 2001
@@ -1705,7 +1705,7 @@
 
 		if ((__s32)when < (__s32)tp->rttvar)
 			when = tp->rttvar;
-		tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, when);
+		tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, min(when, TCP_RTO_MAX));
 	}
 }
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: ip_conntrack locks up hard on 2.4.0 after about 10 hours

[Rusty Russell]

In message <3A573BD2.C7F7771F@voicenet.com> you write:
> It seems that for one reason or another, ip_conntrack totally locks (not
> removeable) after about 10 hours of continued use.  All i found were
> these messages in my dmesg output

What was the contents of /proc/net/ip_conntrack?

Being unremovable can happen if someone is holding a packet, which the
below fix (by Xuan Baldauf) will often alleviate, but connection
tracking doesn't DROP packets (NAT and packet filtering do).

Hope that helps,
Rusty.
--
http://linux.conf.au The Linux conference Australia needed.

diff -urN -I \$.*\$ -X /tmp/kerndiff.RnRDbE --minimal linux-2.4.0-test13-3/net/ipv4/ip_input.c working-2.4.0-test13-3/net/ipv4/ip_input.c
--- linux-2.4.0-test13-3/net/ipv4/ip_input.c	Tue Dec 12 14:28:06 2000
+++ working-2.4.0-test13-3/net/ipv4/ip_input.c	Mon Dec 18 17:07:06 2000
@@ -225,6 +225,13 @@
 	nf_debug_ip_local_deliver(skb);
 #endif /*CONFIG_NETFILTER_DEBUG*/
 
+#ifdef CONFIG_NETFILTER
+	/* Free reference early: we don't need it any more, and it may
+           hold ip_conntrack module loaded indefinitely. */
+	nf_conntrack_put(skb->nfct);
+	skb->nfct = NULL;
+#endif /*CONFIG_NETFILTER*/
+
         /* Point into the IP datagram, just past the header. */
         skb->h.raw = skb->nh.raw + iph->ihl*4;
 
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/