[SERVER-3477] Signal 11 crash Created: 26/Jul/11  Updated: 30/Aug/11  Resolved: 30/Aug/11

Status: Closed
Project: Core Server
Component/s: Sharding, Stability
Affects Version/s: 1.8.2
Fix Version/s: None

Type: Bug Priority: Critical - P2
Reporter: Grégoire Seux Assignee: Eliot Horowitz (Inactive)
Resolution: Duplicate Votes: 0
Labels: mongos
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Linux 2.6.18-238.19.1.el5 #1 SMP Fri Jul 15 07:31:24 EDT 2011 x86_64 x86_64 x86_64 GNU/Linux
CentOS 5


Attachments: Text File issue.log     File mongos_crash_last10min.log.gz    
Issue Links:
Depends
Operating System: Linux
Participants:

 Description   

The mongos crashed with a signal 11 after a lot of "killing old cursor".

I can see a
MessagingPort say send() errno:104 Connection reset by peer CLIENT:52821
Tue Jul 26 10:51:10 [conn177] DBException in process: socket exception
Tue Jul 26 10:51:10 [conn177] MessagingPort say send() errno:32 Broken pipe CLIENT:52821
Tue Jul 26 10:51:10 [conn177] unclean socket shutdown from: CLIENT:52821

just before the crash (see attachment)



 Comments   
Comment by Eliot Horowitz (Inactive) [ 30/Aug/11 ]

see SERVER-3002

Comment by Grégoire Seux [ 03/Aug/11 ]

Ok now I know what may be responsible for this bug : background index building.

My use case is the following :
in the beginning of the operation I create an index using the background option (the collection is empty or so)
for each batch of operation I ensure this index exists
the mongos (why ?) crashes while being under pressure (basically when I modify the field indexed by the background index a lot )

The crash didn't occur before I have discovered the background indexing, and now that I don"t use it anymore, it has not happened again.

Comment by Grégoire Seux [ 01/Aug/11 ]

the last 10 minutes of mongos with -vvv option are in attachment

full log is too heavy for being uploaded on jira, here it is : http://fichiers.ecp.fr/get?k=2ikD4ciTLPPJJGuVMEs

Comment by Scott Hernandez (Inactive) [ 01/Aug/11 ]

It would be good to get more verbose logs (run with --vvvvv) and more than 10 minutes of logs from before the incident.

Are you able to reproduce this error?

Comment by Grégoire Seux [ 26/Jul/11 ]

this bug report follows the thread https://groups.google.com/forum/?hl=fr#!topic/mongodb-user/Ba-23yny9xY

Generated at Thu Feb 08 03:03:10 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.