Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: Apache: Bugs

[Bug 53609] Apache hangs with terminated signal 6

 

 

Apache bugs RSS feed   Index | Next | Previous | View Threaded


bugzilla at apache

Jul 27, 2012, 4:26 AM

Post #1 of 12 (129 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

Eric Covener <covener [at] gmail> changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |NEEDINFO

--- Comment #1 from Eric Covener <covener [at] gmail> ---
What does your error log say? What's the backtrace of the thread that crashed?

"thread apply all bt full" and post the suspicious one.

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 27, 2012, 6:09 AM

Post #2 of 12 (131 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

vinoth <talk2.vino [at] gmail> changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|NEEDINFO |NEW

--- Comment #2 from vinoth <talk2.vino [at] gmail> ---
Apologize previous core info was with other pid. Here is the right one.

The Error Log says:
===================
[Thu Jul 26 22:38:55.033453 2012] [core:notice] [pid 30703:tid 47671878384272]
AH00051: child pid 26692 exit signal Aborted (6), possible coredump in
/piroot/apache2-gdb-dump
apr_table_mergen: key not in ancestor pool of t
[Thu Jul 26 22:43:42.038833 2012] [core:notice] [pid 30703:tid 47671878384272]
AH00051: child pid 6351 exit signal Aborted (6), possible coredump in
/piroot/apache2-gdb-dump
apr_table_mergen: key not in ancestor pool of t


Attached the core dump of thread apply all bt full.

In the dump we removed the sql statements and the client Ipaddr.

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 27, 2012, 6:11 AM

Post #3 of 12 (129 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

--- Comment #3 from vinoth <talk2.vino [at] gmail> ---
Created attachment 29121
--> https://issues.apache.org/bugzilla/attachment.cgi?id=29121&action=edit
Its a complete core dump

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 27, 2012, 6:17 AM

Post #4 of 12 (131 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

--- Comment #4 from Eric Covener <covener [at] gmail> ---
#5 0x00002b5b793d9d2e in apr_table_mergen (t=0x2aab180d29b0, key=0x48169c
"Connection", val=0x481704 "Keep-Alive") at tables/apr_tables.c:739
next_elt = <value optimized out>
end_elt = <value optimized out>
checksum = <value optimized out>
hash = <value optimized out>
#6 0x0000000000468898 in ap_set_keepalive (r=0x2aab181b9a60) at
http_protocol.c:249

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 27, 2012, 7:31 AM

Post #5 of 12 (120 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

Ruediger Pluem <rpluem [at] apache> changed:

What |Removed |Added
----------------------------------------------------------------------------
Component|mpm_worker |APR
Version|2.4.2 |HEAD
Product|Apache httpd-2 |APR

--- Comment #5 from Ruediger Pluem <rpluem [at] apache> ---
I guess this is a bug in APR:

Can you please test the following patch:

Index: tables/apr_tables.c
===================================================================
--- tables/apr_tables.c (revision 1366342)
+++ tables/apr_tables.c (working copy)
@@ -730,15 +730,18 @@
apr_table_entry_t *next_elt;
apr_table_entry_t *end_elt;
apr_uint32_t checksum;
+ apr_pool_t *pool;
int hash;

#if APR_POOL_DEBUG
{
- if (!apr_pool_is_ancestor(apr_pool_find(key), t->a.pool)) {
+ pool = apr_pool_find(key);
+ if ((pool != key) && (!apr_pool_is_ancestor(pool, t->a.pool))) {
fprintf(stderr, "apr_table_mergen: key not in ancestor pool of
t\n");
abort();
}
- if (!apr_pool_is_ancestor(apr_pool_find(val), t->a.pool)) {
+ pool = apr_pool_find(val);
+ if ((pool != val) && (!apr_pool_is_ancestor(pool, t->a.pool))) {
fprintf(stderr, "apr_table_mergen: val not in ancestor pool of
t\n");
abort();
}

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 27, 2012, 10:08 PM

Post #6 of 12 (123 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

--- Comment #6 from vinoth <talk2.vino [at] gmail> ---
Thanks for the patch ruediger. I started the test with the patch against our
module.

I keep you posted if its gets solved our issue.

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 28, 2012, 12:46 AM

Post #7 of 12 (123 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

--- Comment #7 from vinoth <talk2.vino [at] gmail> ---
we again hit with the same error in the patch too while testing. It is not
happened immediately. It occurs during one to two hours of our test and makes
the apache to hang at one stage.

Please let me know whats the real issue with APR table mergen/addn functions.
Why its is behave in that manner when he hit with higher concurrency (100).

Any help to overcome is greatly appreciated.

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 28, 2012, 12:48 AM

Post #8 of 12 (122 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

--- Comment #8 from vinoth <talk2.vino [at] gmail> ---
Created attachment 29123
--> https://issues.apache.org/bugzilla/attachment.cgi?id=29123&action=edit
Error log of the patch apr_table.c

Apache2 error logs

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 28, 2012, 9:37 PM

Post #9 of 12 (121 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

--- Comment #9 from vinoth <talk2.vino [at] gmail> ---
Found one interesting in the error log .apache2 gets restarted every time when
the worker getting shutdown when it caught SIGTERM. While worker is getting
shutting down we don't have any core dump associated with that.

I set the maxreqworker is 150. The concurrency hit to the apache is always less
than the maxreqworker config.

Is this the expected behaviour?

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 30, 2012, 2:11 AM

Post #10 of 12 (104 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

--- Comment #10 from Ruediger Pluem <rpluem [at] apache> ---
(In reply to comment #7)
> we again hit with the same error in the patch too while testing. It is not
> happened immediately. It occurs during one to two hours of our test and
> makes the apache to hang at one stage.

I cannot see that you hit the same error again. You still have segfaults /
aborts, but you don't have the "key not in ancestor pool of " any longer in the
log. This means that the patch works and resolved one of your issues.

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 30, 2012, 5:27 AM

Post #11 of 12 (105 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

--- Comment #11 from Ruediger Pluem <rpluem [at] apache> ---
(In reply to comment #10)
> (In reply to comment #7)
> > we again hit with the same error in the patch too while testing. It is not
> > happened immediately. It occurs during one to two hours of our test and
> > makes the apache to hang at one stage.
>
> I cannot see that you hit the same error again. You still have segfaults /
> aborts, but you don't have the "key not in ancestor pool of " any longer in
> the log. This means that the patch works and resolved one of your issues.

The patch was committed as r1367050.

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd


bugzilla at apache

Jul 30, 2012, 9:31 AM

Post #12 of 12 (110 views)
Permalink
[Bug 53609] Apache hangs with terminated signal 6 [In reply to]

https://issues.apache.org/bugzilla/show_bug.cgi?id=53609

--- Comment #12 from vinoth <talk2.vino [at] gmail> ---
We see those error "key not in ancestor pool of t" in the attached Error logs

===============
[Fri Jul 27 21:36:59.029498 2012] [core:notice] [pid 449:tid 47538503809680]
AH00094: Command line: '/usr/local/apache2/bin/httpd'
apr_table_mergen: key not in ancestor pool of t
[Fri Jul 27 22:29:38.067056 2012] [core:notice] [pid 449:tid 47538503809680]
AH00051: child pid 1083 exit signal Aborted (6), possible coredump in
/piroot/apache2-gdb-dump
apr_table_mergen: key not in ancestor pool of t
[Fri Jul 27 22:50:13.089987 2012] [core:notice] [pid 449:tid 47538503809680]
AH00051: child pid 23412 exit signal Aborted (6), possible coredump in
/piroot/apache2-gdb-dump
apr_table_addn: val not in ancestor pool of t
[Fri Jul 27 22:50:22.091015 2012] [core:notice] [pid 449:tid 47538503809680]
AH00051: child pid 32280 exit signal Aborted (6), possible coredump in
/piroot/apache2-gdb-dump
*** glibc detected *** /usr/local/apache2/bin/httpd: double free or corruption
(!prev): 0x00002aab90008b30 ***
*** glibc detected *** /usr/local/apache2/bin/httpd: double free or corruption
(!prev): 0x00002aab90008b30 ***
*** glibc detected *** /usr/local/apache2/bin/httpd: double free or corruption
(!prev): 0x00002aab90008b30 ***
*** glibc detected *** /usr/local/apache2/bin/httpd: double free or corruption
(!prev): 0x00002aab90008b30 ***
*** glibc detected *** /usr/local/apache2/bin/httpd: double free or corruption
(!prev): 0x00002aab90008b30 ***
*** glibc detected *** /usr/local/apache2/bin/httpd: double free or corruption
(!prev): 0x00002aab90008b30 ***
[Fri Jul 27 22:50:33.091944 2012] [core:notice] [pid 449:tid 47538503809680]
AH00051: child pid 32614 exit signal Segmentation fault (11), possible coredump
in /piroot/apache2-gdb-dump
apr_table_addn: val not in ancestor pool of t
[Fri Jul 27 22:50:54.093086 2012] [core:notice] [pid 449:tid 47538503809680]
AH00051: child pid 525 exit signal Aborted (6), possible coredump in
/piroot/apache2-gdb-dump
apr_table_mergen: key not in ancestor pool of t
[Fri Jul 27 23:19:04.121667 2012] [core:notice] [pid 449:tid 47538503809680]
AH00051: child pid 959 exit signal Aborted (6), possible coredump in
/piroot/apache2-gdb-dump
apr_table_addn: val not in ancestor pool of t
[Fri Jul 27 23:35:08.139140 2012] [core:notice] [pid 449:tid 47538503809680]
AH00051: child pid 13036 exit signal Aborted (6), possible coredump in
/piroot/apache2-gdb-dump

===============

Today we tried out the prefork mpm for the test instead of worker mpm. In that
we dont see any aborts and seg faults in the test and also the error "key not
in ancestor pool of t".

Is there is anything associated with the issue in worker mpm.

When we hit with the issue while in worker mpm the backtrace always say
something that releated to worker.c file

Is there any other information that you need from my side.

--
You are receiving this mail because:
You are the assignee for the bug.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe [at] httpd
For additional commands, e-mail: bugs-help [at] httpd

Apache bugs RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.