Nginx1.0.0 have a kernel bug

OS is redhat5.5 ,nginx 1.0.0 ,libunwind-0.99…
The /var/log/message frequently appear the follow messages:
Jun 13 10:20:31 localhost kernel: nginx[14431]: segfault at 2a2 ip
00007fa6b3a05eb6 sp 00007fff337d87e0 error 4 in
libunwind.so.7.0.0[7fa6b39fb000+f000]
Jun 13 16:01:09 localhost kernel: nginx[13794]: segfault at 2a2 ip
00007fa6b3a05eb6 sp 00007fff337d87a0 error 4 in
libunwind.so.7.0.0[7fa6b39fb000+f000]
Jun 13 17:55:00 localhost kernel: nginx[13808]: segfault at 2a2 ip
00007fa6b3a05eb6 sp 00007fff337d86e0 error 4 in
libunwind.so.7.0.0[7fa6b39fb000+f000]
Jun 14 01:13:48 localhost kernel: nginx[14592]: segfault at 2a2 ip
00007fa6b3a05eb6 sp 00007fff337d87a0 error 4 in
libunwind.so.7.0.0[7fa6b39fb000+f000]
Jun 14 01:47:13 localhost kernel: nginx[13153]: segfault at 2a2 ip
00007fa6b3a05eb6 sp 00007fff337d87e0 error 4 in
libunwind.so.7.0.0[7fa6b39fb000+f000]
Jun 14 09:09:28 localhost kernel: nginx[28921]: segfault at 8008 ip
00007fa6b3a05eb6 sp 00007fff337d8c60 error 4 in
libunwind.so.7.0.0[7fa6b39fb000+f000]
Jun 14 15:47:33 localhost kernel: nginx[21295]: segfault at 2a2 ip
00007fde9e837eb6 sp 00007fffc48594a0 error 4 in
libunwind.so.7.0.0[7fde9e82d000+f000]

Can you help me solve it?

Posted at Nginx Forum:
http://forum.nginx.org/read.php?2,207390,207390#msg-207390

this is not a bug in the kernel, but the nginx is trying to access
some memory that it is not allowed to access.

did you add any modules? what is libunwind used for?
can you send the output of nginx -V

i think you should probably first find out which of all your
additional modules is linking to libunwind.so.

then i would guess there must be a bug either in that module or in the
libunwind.

if you compile nginx without any add-module i think it should not link
to libunwind, so this must come from one of your via --add-module
added modules.

Thanks!
The output of nginx -V is:
[[email protected] ~]# /web/software/nginx/sbin/nginx -V
nginx: nginx version: nginx/1.0.0
nginx: TLS SNI support enabled
nginx: configure arguments: --prefix=/web/software/nginx
–with-google_perftools_module --user=www --group=www
–prefix=/web/software/nginx --with-http_stub_status_module
–with-http_ssl_module --with-pcre --add-module=…/ngx_cache_purge-1.3
–add-module=…/ngx_http_redis-0.3.2
–add-module=…/kkung-nginx-static-etags-ce0fdd4
–add-module=…/agentzh-memc-nginx-module-25454a8
–add-module=…/ngx_http_upstream_keepalive-2ee28064a04a
–add-module=…/simpl-ngx_http_set_hash-cfdd587
–add-module=…/simpl-ngx_devel_kit-7dbb4ce
–with-ld-opt=’-L/web/software/openssl/lib/ -L/web/software/zlib/lib/’

Posted at Nginx Forum:
http://forum.nginx.org/read.php?2,207390,207397#msg-207397

I think the with-google_perftools_module will link to libunwind,but i
don’t known how do prove and solve it.
because google_perftools need libunwind under the X86_64 system .

Posted at Nginx Forum:
http://forum.nginx.org/read.php?2,207390,207431#msg-207431

if you want to be sure, recompile the nginx without google_perftools,
then you do ldd on the nginx binary and see if it still links to
libunwind.

once you are sure its coming from this module, you can start trying to
find bugreports of this problem or try other versions of either the
library or the nginx module.

replay Wrote:
i try to update nginx to 1.0.4 ,but i meet another problem,when i use
nginx 1.0.1 or 1.0.4 after about10 hours or more ,the nginx don’t
working i cant open the web or the nginx_status information,Surprising
that the nginx process still exist .if happen this ,i must restart
nginx.when i back to use nginx 1.0.0, the problem willn’t appear.

library or the nginx module.


nginx mailing list
[email protected]
http://nginx.org/mailman/listinfo/nginx

Posted at Nginx Forum:
http://forum.nginx.org/read.php?2,207390,207440#msg-207440

This forum is not affiliated to the Ruby language, Ruby on Rails framework, nor any Ruby applications discussed here.

| Privacy Policy | Terms of Service | Remote Ruby Jobs