Bug #57556 Rolling restart of added ndbmtd never completes
Submitted: 19 Oct 2010 9:07
Reporter: Thomas Nielsen Email Updates:
Status: Verified Impact on me:
None 
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S2 (Serious)
Version:mysql-5.1-telco-7.1 OS:Linux (sles10)
Assigned to: CPU Architecture:Any
Tags: 7.1.8
Triage: Triaged: D2 (Serious) / R6 (Needs Assessment) / E6 (Needs Assessment)

[19 Oct 2010 9:07] Thomas Nielsen
Description:
Rolling restart of a previously added ndbmtd, sometimes hang during subsequent rolling restart when adding more processes

How to repeat:
create and start cluster
  #1 mgmd
  #2 ndbmtd
  #3 ndbmtd
  #4 mysqld
Add second mgmd
  #5 mgmd
Rolling restart #1-4, then start #5
Add two ndbmtds
  #6 ndbmtd
  #7 ndbmtd
Rolling restart #1-5, then start #6+7
Create nodegroup 6,7
Add mysqld
  #8 mysqld
Rolling restart #1-7, then start #8  <-- sometimes hangs at restart of #6
[19 Oct 2010 9:08] Thomas Nielsen
ndb_error_reporter-like tarball of available files relating to the error

Attachment: ndb_error_reporter_20101015063434.tar.bz2 (application/bzip2, text), 6.91 KiB.

[19 Oct 2010 9:16] Thomas Nielsen
Last repro step "Rolling restart #1-7", is restarting mgmds first, then ndb(mt)ds, then mysqlds. Detailed restart order is as follows: 1,5,2,3,6,7,4