Bug #32672 "Start of Master cluster failed" for all NDB tests after a "ndb_restore" failure

• Submitted: 23 Nov 2007 17:15
• Modified: 14 Apr 2010 13:46
• Reporter: Joerg Bruehe
• Status: Can't repeat
• Category: MySQL Cluster: Cluster (NDB) storage engine
• Severity: S3 (Non-critical)
• Version: mysql-5.1-telco-6.3
• OS: Other (Solaris 8, x86; Linux, x86_64)
• Assigned to: Magnus Blåudd
• CPU Architecture: Any
• Tags: 5.0.51 + 6.3.17

[23 Nov 2007 17:15] Joerg Bruehe

Description:
Found in a buld of version 5.0.51.

Test "ndb_restore_different_endian_data" fails (bug#31453).
This happens on all platforms, but does not cause further problems -
except for Solaris 8 (x86) in "PS protocol" mode.

Here, all subsequent tests involving NDB fail with the same symptom:

=====
ndb_restore_print              [ fail ]

ERROR: Start of Master cluster failed

Stopping All Servers
Restoring snapshot of databases
Resuming Tests
=====

Bug#31453 happens in all runs (debug, normal, PS), but this NDB start problem happens only in the "PS" run;
bug#31453 also happened in 5.0.50, but there the subsequent NDB start problem did not occur.

Affected tests:
ndb_restore_print
ndb_single_user
ndb_subquery
ndb_transaction
ndb_trigger
ndb_truncate
ndb_types
ndb_update
ps_7ndb
rpl_ndb_innodb_trans
strict_autoinc_5ndb

How to repeat:
Occurred during the build test run.

[8 Sep 2008 10:27] Joerg Bruehe

In the build of Cluster-6.3.17, I have the symptom
   ERROR: Start of Master cluster failed
occur 54 times,

all on the same platform (Linux, x86_64, RPM build for SuSE 9),

in all three build runs
(different configurations: "cluster-gpl", "cluster-com", "cluster-com-pro").

Opposed to the original report here, there is no clear pattern in the preceding tests:
Mostly, it happens after one or more preceding "pass",
but also after "skipped", "disabled", and "fail".

I know that this description does not provide much info,
but as the maintainer of "mysql-test-run.pl" is a member of the cluster team,
it might be possible to automatically collect some info after such a failure to gain insight into the causes ?