os-linux/5301: Apache crashes Linux box when serving a busy site.

[Zh...@hyperreal.org, Hongsheng <ge...@netfirm.net>]

>Number:         5301
>Category:       os-linux
>Synopsis:       Apache crashes Linux box when serving a busy site.
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    apache
>State:          open
>Class:          sw-bug
>Submitter-Id:   apache
>Arrival-Date:   Sat Nov 13 07:20:00 PST 1999
>Last-Modified:
>Originator:     george@netfirm.net
>Organization:
apache
>Release:        1.3.9
>Environment:
Linux Mandrake 6.1, kernel 2.2.13-7mdk, running on a PII/350 w/ 384M RAM. pgcc 1.1.3-3mdk.
>Description:
I'm running a site with many visitors at the same time, normally there are about 80 to 120 clients at the same time, and the visitors often come in and stay for a short time, then leave. The active connections remain in 80 to 120, but there are lots of SYN, CLOSE, FIN_WAIT1 ports left. Apache starts working fine, but after some time (not a regular value), apache will crash the whole Linux box with the following error messages logged:

kernel: Unable to handle kernel paging request at virtual address 3afe02d0
kernel: current->tss.cr3 = 09781000, %cr3 =09781000
kernel: *pde = 00000000
kernel: Oops: 0002
kernel: CPU:    0
kernel: EIP:    0010:[alloc_skb+200/240]
kernel: EFLAGS: 00010202
kernel: eax: 3afe02d0   ebx: cf0dc3c0   ecx: cd258800   edx: 00000000
kernel: esi: 000002d0   edi: 00000015   ebp: c99f5e74   esp: c99f5e04
kernel: ds: 0018   es: 0018   ss: 0018
kernel: Process httpd (pid: 440, process nr: 44, stackpage=c99f5000)
kernel: Stack: c8929cc0 c015ae6f 000002c8 00000015 00002180 c01722b7 c8929cc0 000002c8
kernel:        00000000 00000015 c99f5f00 c8929cc0 c0182388 00000218 00000001 c99f4000
kernel:        c99f5e70 c8929d70 00000000 40207305 00007cfb 00000430 00000218 00000000
kernel: Call Trace: [sock_wmalloc+35/72] [tcp_do_sendmsg+955/2016] [inet_sendmsg+0/144] [tcp_v4_sendmsg+92/104] [inet_sendmsg+131/144] [sock_sendmsg+136/172] [inet_sendmsg+0/144]
kernel:        [sock_readv_writev+128/140] [do_readv_writev+321/472] [old_mmap+195/260] [old_mmap+212/260] [sys_writev+124/176] [system_call+52/56]
kernel: Code: c7 00 01 00 00 00 89 d8 eb 1a89 f6 53 a1 c4 48 27 c0 50 e8


>How-To-Repeat:
I'm not sure how to replay this problem, the following maybe helpful:

Create a site, and try to access it with many simultaneous requests, then finish these requests quickly, then start a new turn of simultaneous requests, and so on, after some time, normally 1/2 to 2 hours, apache may crashes the system.
>Fix:
I have no idea on this.
>Audit-Trail:
>Unformatted:
[In order for any reply to be added to the PR database, you need]
[to include <ap...@Apache.Org> in the Cc line and make sure the]
[subject line starts with the report component and number, with ]
[or without any 'Re:' prefixes (such as "general/1098:" or      ]
["Re: general/1098:").  If the subject doesn't match this       ]
[pattern, your message will be misfiled and ignored.  The       ]
["apbugs" address is not added to the Cc line of messages from  ]
[the database automatically because of the potential for mail   ]
[loops.  If you do not include this Cc, your reply may be ig-   ]
[nored unless you are responding to an explicit request from a  ]
[developer.  Reply only with text; DO NOT SEND ATTACHMENTS!     ]