MySQL Bugs: #30544: "ERROR: Cluster Master was not installed ok" => several tests fail

Bug #30544	"ERROR: Cluster Master was not installed ok" => several tests fail
Submitted:	21 Aug 2007 19:56	Modified:	4 Sep 2008 17:23
Reporter:	Joerg Bruehe	Email Updates:
Status:	Verified	Impact on me:	None
Category:	MySQL Cluster: Cluster (NDB) storage engine	Severity:	S1 (Critical)
Version:	mysql-5.0	OS:	Any
Assigned to:		CPU Architecture:	Any
Tags:	5.0 and up, disabled, incl CGE

Description:
Occurred during the build of 5.1.21-beta.

The bug is specific to one platform: Mac OS X, PowerPC, 32 bit.
In the 64 bit variant, the tests did not progress so far,
so I cannot tell whether it passed or failed there.
On all other platforms which include "cluster", these tests passed
(including OS X on an x86 CPU).

Symptom:
=====
binlog.binlog_multi_engine     [ fail ]

ERROR: Cluster Master was not installed ok
=====

Affected tests (same symptom:
binlog.binlog_multi_engine
ndb.ndb_alter_table3
ndb.ndbapi
ndb.ndb_dd_dump
ndb.ndb_limit
ndb.ndb_partition_error
ndb.strict_autoinc_5ndb
rpl_ndb.rpl_ndb_auto_inc
rpl_ndb.rpl_ndb_commit_afterflush
rpl_ndb.rpl_ndb_dd_basic
rpl_ndb.rpl_ndb_idempotent
rpl_ndb.rpl_ndb_insert_ignore
rpl_ndb.rpl_ndb_log
rpl_ndb.rpl_ndb_multi_update3
rpl_ndb.rpl_ndb_sp006

This is the log of test suite start:
=====
cd mysql-test ; MTR_BUILD_THREAD=auto \
/usr/bin/perl ./mysql-test-run.pl --comment=ps+rowrepl+NDB --force --timer \
--ps-protocol --mysqld=--binlog-format=row
Requesting build thread... got 200
Logging: ./mysql-test-run.pl --comment=ps+rowrepl+NDB --force --timer --ps-protocol --mysqld=--binlog-format=row
070817  2:44:18 [Warning] Setting lower_case_table_names=2 because file system for /Users/mysqldev/tmp-200708161639-5.1.21-beta-4346/osx-tiger-ppc/test/mysql-5.1.21-beta-osx10.4-powerpc/share/mysql/english/ is case insensitive
MySQL Version 5.1.21

##############################################################################
# ps+rowrepl+NDB
##############################################################################

Using binlog format 'row'
Using ndbcluster when necessary, mysqld supports it
Setting mysqld to support SSL connections
Using MTR_BUILD_THREAD      = 200
Using MASTER_MYPORT         = 12000
Using MASTER_MYPORT1        = 12001
Using SLAVE_MYPORT          = 12002
Using SLAVE_MYPORT1         = 12003
Using SLAVE_MYPORT2         = 12004
Using NDBCLUSTER_PORT       = 12005
Using NDBCLUSTER_PORT_SLAVE = 12006
Using IM_PORT               = 12007
Using IM_MYSQLD1_PORT       = 12008
Using IM_MYSQLD2_PORT       = 12009
Killing Possible Leftover Processes
mysql-test-run: WARNING: Found non pid file master-slow.log in /Users/mysqldev/tmp-200708161639-5.1.21-beta-4346/osx-tiger-ppc/test/mysql-5.1.21-beta-osx10.4-powerpc/mysql-test/var/run
Removing Stale Files
Creating Directories
Installing Master Database
Installing Master Database
Installing Slave1 Database
Installing Master Cluster
Installing Slave Cluster
ndbcluster_install of Master failed
ndbcluster_install of Slave failed
Saving snapshot of installed databases
=======================================================

How to repeat:
Found when running the full test suite.

What system(s) did this happen on? 

Is there any logs that show why it was not installed? 

Have you repeated this? (ie. Does this happen every time? Every other time? Just once or what?) 

Can you make these systems available for debugging if this is repeatable?

Thanks!

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

We discussed this somewhat during the Heidelberg Dev meeting, AFAIR,
but obviously didn't add info here. :-(

To add a current data point:
During the build of Cluster-6.3.17,
summed across all platforms and configurations,
I had this symptom occur in 110 test suite runs
(affecting more than 3,400 times - yes, three thousand !).

Typically, this was in RPM builds on Linux.

So this is still an issue,
it occurs quite frequently,
it happens on the build hosts,
and the trees there are archived
(but will be removed soon, for lack of disk space).

In 5.0.72, we had this symptom on Linux (SuSE 10 / x86_64).

In 5.0.77, I have it on AIX 5.3 (PPC, 32 bit).

I *guess* this is just some assumption about the installation which is sometimes invalidated by machine conditions, but the message is simply too vague to give me any hint.

See also: BUG#47248.