Bug #49219 ndb_mgm segfaults when receiving command after ndb_mgmd restart
Submitted: 30 Nov 2009 17:34 Modified: 3 Mar 2010 12:27
Reporter: Erik Ljungstrom Email Updates:
Status: Closed Impact on me:
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S3 (Non-critical)
Version:mysql-5.1-telco-7.0 OS:Linux (CentOS 5.3)
Assigned to: Magnus Blåudd CPU Architecture:Any
Tags: 7.0.9, NDB_MGM

[30 Nov 2009 17:34] Erik Ljungstrom
ndb_mgm segfaults when it's lost its connection to ndb_mgmd, and another mgmd process has been spawned without restarting the ndb_mgm session.


Thread 1 (process 8437):
#0  0x0000003cb52705c4 in malloc_consolidate () from /lib64/libc.so.6
#1  0x0000003cb5272a1a in _int_malloc () from /lib64/libc.so.6
#2  0x0000003cb5274bee in malloc () from /lib64/libc.so.6
#3  0x00000000004f29d0 in my_malloc (size=648, my_flags=648) at my_malloc.c:37
#4  0x00000000004684a5 in ndb_mgm_create_handle () at mgmapi.cpp:187
#5  0x0000000000461a45 in CommandInterpreter::connect (this=0x3cb55529e0, interactive=true) at CommandInterpreter.cpp:960
#6  0x0000000000466b13 in CommandInterpreter::execute_impl (this=0x1a1dabf0, _line=0x1a207d20 "show", interactive=true)
    at CommandInterpreter.cpp:1184
#7  0x0000000000466d80 in CommandInterpreter::execute (this=0x3cb55529e0, _line=0x288 <Address 0x288 out of bounds>, 
    _try_reconnect=<value optimized out>, interactive=144, error=0x0) at CommandInterpreter.cpp:1096
#8  0x0000000000460fce in main (argc=0, argv=0x1a1daad0) at main.cpp:130

Expected: ndb_mgm reconnects and issues command

How to repeat:
Start an ndb_mgm process in session A and run show.
Start another ndb_mgm process in session B.
Issue 'shutdown' in session B.
Start ndb_mgmd again, and issue 'show' in session A:
ndb_mgm> show
Segmentation fault (core dumped)
[30 Nov 2009 17:43] MySQL Verification Team
Thank you for submitting this bug report. 
Verified as described.
[17 Feb 2010 14:10] Magnus Blåudd
Can't get it to crash but running in valgrind shows several warnings while reconnecting after SIGPIPE.
[17 Feb 2010 14:13] Magnus Blåudd
Occurs also in 6.3
[17 Feb 2010 15:28] Bugs System
A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

[3 Mar 2010 10:17] Magnus Blåudd
Pushed to 6.3.33, 7.0.14 and 7.1.2-beta
[3 Mar 2010 12:27] Jon Stephens
Documented bugfix in the NDB-6.3.33, 7.0.14, and 7.1.2 changelogs, as follows:

      Issuing a command in the ndb_mgm client after it had lost its 
      connection to the management server could cause the client to crash.