Bug #27641 ndb_mgmd does core dump with wrong hostname settings

• Submitted: 4 Apr 2007 9:46
• Modified: 16 Apr 2007 10:58
• Reporter: Oli Sennhauser
• Status: Can't repeat
• Category: MySQL Cluster: Cluster (NDB) storage engine
• Severity: S3 (Non-critical)
• Version: 5.1.16
• OS: Solaris (SunOS BTO3E05DBS0S 5.10 Generic_118855-33 i86pc i386 i86pc)
• CPU Architecture: Any

[4 Apr 2007 9:46] Oli Sennhauser

Description:
after reconfiguration of config.ini the ndb_mgmd was not able to start again and did a core dump.
after back reconfiguration it worked again. I do agree, that configuration is wrong but it should not crash anyway...

How to repeat:
see attached config.ini file

Suggested fix:
quit with proper message and not with core dump

[5 Apr 2007 23:43] Hartmut Holzgraefe

Can't reproduce on linux/x86 with 5.1.16

For me it prints

  Cluster configuration warning:
    arbitrator with id 1 and db node with id 10 on same host 172.16.3.15
    arbitrator with id 2 and db node with id 11 on same host 172.16.3.18
    Running arbitrator on the same host as a database node may
    cause complete cluster shutdown in case of host failure.
  Unable to obtain requested nodeid: Connection done from wrong host ip .

followed by

  Aborted (core dumped)

The core dump is due to --core-file being enabled by default
(at least in debug builds, didn't test non-debug yet)

Can you try to provide a backtrace from the core dump?

[11 Apr 2007 9:51] Oli Sennhauser

I am not on site anymore