- To: outages-list@eskimo.com, linux-kernel@vger.rutgers.edu, Alan Cox <alan@lxorguk.ukuu.org.uk>
- Subject: WWW Crash
- From: Robert Dinse <nanook@eskimo.com>
- Date: Tue, 6 Jun 2000 21:34:55 -0700 (PDT)
- Resent-Date: Tue, 6 Jun 2000 21:34:59 -0700
- Resent-From: outages-list@eskimo.com
- Resent-Message-ID: <"LVN7C.0.n6.p3TFv"@mx1>
- Resent-Sender: outages-list-request@eskimo.com
- Prev by Date: WWW patched...
- Next by Date: WWW freeze 6/7 7:40pm
- Prev by thread: WWW Crash
- Next by thread: Re: WWW Crash
- Index(es):
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
WWW Crash
Alan, I applied the following patch:
--- linux/net/ipv4/tcp_output.c.orig Tue May 30 23:25:13 2000
+++ linux/net/ipv4/tcp_output.c Thu Jun 1 22:39:25 2000
@@ -596,6 +596,7 @@
(skb->len < (cur_mss >> 1)) &&
(skb->next != tp->send_head) &&
(skb->next != (struct sk_buff *)&sk->write_queue) &&
+ (skb->next != tp->retrans_head) &&
(sysctl_tcp_retrans_collapse != 0))
tcp_retrans_try_collapse(sk, skb, cur_mss);
... to 2.2.15, otherwise stock.
Running on quad CPU SS-10, Ross RTK-625 Hypersparc, after about an hour
we got a couple of OOPS in apached and then the machine hung.
It wasn't doing this without the patch, though without the patch I did get
an occasional OOPS in swapper followed by machine death.
We've rebooted with the patch still in place and will allow it to run a
while longer to find out if this patch is the casue or not. I captured the
OOP's with a digital camera and will post those after I have time to transcribe
but may be a day or two.