[SERVER-10331] Mongodb SEGFAULTS with a null pointer Created: 25/Jul/13  Updated: 02/Aug/13  Resolved: 25/Jul/13

Status: Closed
Project: Core Server
Component/s: Concurrency, Stability
Affects Version/s: 2.2.3
Fix Version/s: None

Type: Bug Priority: Blocker - P1
Reporter: Jim Assignee: Michael Grundy
Resolution: Duplicate Votes: 0
Labels: crash
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Centos 6.2, pymongo-2.1.1-1, pulp-2.1.3-1


Attachments: File mongodb.log.2013-07-24T15-22-37    
Issue Links:
Duplicate
duplicates SERVER-9014 Mongod and mongos crash induced by ma... Closed
Backwards Compatibility: Fully Compatible
Operating System: Linux
Steps To Reproduce:
  1. service mongod status
    mongod dead but pid file exists
  2. service mongod restart
    Stopping mongod: [FAILED]
    Starting mongod: [ OK ]
  3. service mongod status
    mongod (pid 30930) is running...
  4. pulp-manage-db
    Beginning database migrations.
    Migration package pulp.server.db.migrations is up to date at version 4
    Migration package pulp_rpm.migrations is up to date at version 7
    Database migrations complete.
    Loading content types.
    Content types loaded.
  5. service httpd restart
    Stopping httpd: [ OK ]
    Starting httpd: [ OK ]
  6. service mongod status
    mongod dead but pid file exists

-------------------------------------
Method 2)

  1. mongo pulp_database < echo "db.queued_calls.drop();"
    MongoDB shell version: 2.2.3
    connecting to: pulp_database
    false
    Wed Jul 24 10:25:24 mongo got signal 11 (Segmentation fault), stack trace:

Wed Jul 24 10:25:24 0x59e416 0x47adbd 0x3a41032900
mongo(_ZN5mongo15printStackTraceERSo+0x26) [0x59e416]
mongo(_Z12quitAbruptlyi+0x31d) [0x47adbd]
/lib64/libc.so.6() [0x3a41032900]

Participants:

 Description   

I am trying to use pulp, which a program that manages package repositories, and uses mongodb as a backend. However, it has been causing mongodb to segfault. https://bugzilla.redhat.com/show_bug.cgi?id=986070 http://pastebin.com/VgfHZj0q (the file being piped in just contains "db.queued_calls.drop();" ). The second link in particular seems to point the finger at mongo.

Pulp has a db purging call queued, and on its startup it segfaults mongodb, which itself complains about a null pointer, and what appears to a threading problem (as soon as anything thread but the first does anything it gets said null pointer)



 Comments   
Comment by Jim [ 25/Jul/13 ]

Thanks for the help; its good to have confirmation that this is a known issue that was indeed fixed.

Comment by Jim [ 25/Jul/13 ]

Also, after opening this case I found this bug: https://jira.mongodb.org/browse/SERVER-9014, of which this looks as though it may be a duplicate.

I upgraded one of my servers to 2.4.5, and it hasnt segfaulted in the ~20 min since...

Comment by Michael Grundy [ 25/Jul/13 ]

Hi Jim -

Glad you found the pattern, that is fixed in 2.4.2, but we recommend you update to the latest, which is currently 2.4.5 and available at http://downloads.mongodb.org

Thanks!
Mike

Comment by Jim [ 25/Jul/13 ]

Full log from a run of this

Comment by Michael Grundy [ 25/Jul/13 ]

Could you post the complete mongod log from your test case? The backtrace looks like it is from the mongo shell, not mongod.

Thanks!
Mike

Generated at Thu Feb 08 03:22:53 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.