MySQL Bugs: #39004: Cluster failed to start during test run (in PB2 and in release build)

Bug #39004	Cluster failed to start during test run (in PB2 and in release build)
Submitted:	25 Aug 2008 11:56	Modified:	29 Oct 2010 7:36
Reporter:	Sven Sandberg	Email Updates:
Status:	Won't fix	Impact on me:	None
Category:	Tests: Cluster	Severity:	S7 (Test Cases)
Version:	5.1	OS:	Any
Assigned to:		CPU Architecture:	Any
Tags:	pb2, sporadic, test failure, widespread

Description:
pushbuild failure:

binlog.binlog_multi_engine               [ fail ]

CURRENT_TEST: rpl_ndb.rpl_ndb_2other
CURRENT_TEST: rpl_ndb.rpl_ndb_innodb_trans
CURRENT_TEST: rpl_ndb.rpl_ndb_innodb_trans
CURRENT_TEST: binlog.binlog_multi_engine

Start of 'mysql_cluster.1' cluster failed

 - saving '/tmp/var-ps_row-101/2/log/binlog.binlog_multi_engine/' to '/tmp/var-ps_row-101/log/binlog.binlog_multi_engine/'

Retrying test, attempt(2/3)...

How to repeat:
https://intranet.mysql.com/secure/pushbuild/showpush.pl?dir=bzr_mysql-5.1-rpl&order=36 sol10-sparc-a/n_mix and ps_row

The bug wasn't for binlog.binlog_multi_engine only but for the group of tests. There were tests before and after that with same error so it is PB/mtr issue.

Setting to "won't fix" since:
 - the error happened for other test cases (so not related to any specific test case);
 - there is no way to debug failures like this one.
Likely a host environment problem or an mtr problem.

The server log seems to have disappeared from pushbuild. Attaching it instead.

server log for 5.1-rpl/36 sol10-sparc-a/n_mix

Attachment: 5.1-rpl_36_sol10-sparc-a_n_mix (application/octet-stream, text), 161.16 KiB.

Re-opening this bug, because this kind of problem happens regularly.

XRef:
http://pb2.norway.sun.com/web.py?template=mysql_show_test_failure&test_failure_id=2356855&...

The very same issue happened 24 times in the build of Cluster-7.1.9
just on one platform: OS X 10.5 (x86).

As this is not in PB2, I change the synopsis.

I do not understand why a problem "Cluster failed to start" is assigned to someone outside the Cluster team, so I reset all these assignments.

As long as this occurs, I need this bug entry so that I can assign this bug number to the respective test failures.

Of course, it would help if the error message would provide more info ...