Bug #14581 ndbd segfaults if it can't find a free node slot to connect to
Submitted: 2 Nov 2005 17:22 Modified: 4 Nov 2005 22:50
Reporter: Tobias Asplund Email Updates:
Status: Closed Impact on me:
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S2 (Serious)
Version: OS:
Assigned to: CPU Architecture:Any

[2 Nov 2005 17:22] Tobias Asplund

NoOfReplicas = 2
DataDir = /var/ndb
FileSystemPath = /var/ndb

Hostname = localhost
Id = 1

Hostname = localhost
Id = 2

Hostname = localhost
Id = 3

How to repeat:
root@eeyore:/var/ndb# /usr/local/mysql/libexec/ndbd
root@eeyore:/var/ndb# /usr/local/mysql/libexec/ndbd
root@eeyore:/var/ndb# /usr/local/mysql/libexec/ndbd
Segmentation fault (core dumped)

root@eeyore:/var/ndb# cat ndb_pid30768_error.log
Current byte-offset of file-pointer is: 568

Time: Wednesday 2 November 2005 - 18:18:33
Status: Permanent error, external action needed
Message: Invalid configuration received from Management Server (Configuration error)
Error: 2350
Error data: Unable to alloc node id
Error object: Could not alloc node id at localhost port 1186: No free node id found for ndbd(NDB).
Program: /usr/local/mysql/libexec/ndbd
Pid: 30768
Trace: <no tracefile>
Version: Version 5.0.15

The error is what we expect, but I don't think a coredump is.
[2 Nov 2005 19:02] Jonathan Miller
I agree with the customer that the NDB should not core.
[4 Nov 2005 22:41] Hartmut Holzgraefe
verified with 5.0.15 but not with a pre-5.0.15 bk build, testing latest 5.0bk now
[4 Nov 2005 22:50] Hartmut Holzgraefe
looks like something that only affects 5.0.15, with latest 5.0bk sources it terminates gracefully with

2005-11-04 23:47:21 [ndbd] INFO     -- Error handler restarting system
2005-11-04 23:47:21 [ndbd] INFO     -- Error handler shutdown completed - aborting

as it did with my older pre-5.0.15 bk build