MySQL Bugs: #34103: create undo file outside datadir fails

Bug #34103	create undo file outside datadir fails
Submitted:	28 Jan 2008 12:58	Modified:	13 Jun 2008 22:35
Reporter:	Johan Andersson	Email Updates:
Status:	No Feedback	Impact on me:	None
Category:	MySQL Cluster: Disk Data	Severity:	S3 (Non-critical)
Version:	5.1.23 ndb 6.3.7	OS:	Linux
Assigned to:		CPU Architecture:	Any
Tags:	cluster disk data, disk data

Description:
Follow on bug 34102... (same config):

I lowered the UNDO_BUFFER_SIZE to 128M (in order to avoid bug 34102)
I want to create the undo_1.dat in a directory other than the Datadir of the data nodes to make use of other disk spindles:

mysql> CREATE LOGFILE GROUP lg_1 ADD UNDOFILE '/data0/johan/mysqlcluster/undo_1.dat' INITIAL_SIZE=4096M UNDO_BUFFER_SIZE=128M ENGINE=ndb;
ERROR 1528 (HY000): Failed to create UNDOFILE
mysql> show errors;
+-------+------+----------------------------------------------------------------------------+
| Level | Code | Message                                                                    |
+-------+------+----------------------------------------------------------------------------+
| Error | 1296 | Got error 1509 'File system error, check if path,permissions etc' from NDB | 
| Error | 1528 | Failed to create UNDOFILE                                                  | 

But I have the permissions set correctly!!!

-bash-3.1$  ls -al  /data0/johan/mysqlcluster
total 951312
drwxrwxrwx 2 root root      4096 Jan 28 13:14 .
drwxrwxrwx 3 root root      4096 Jan 28 13:14 ..
-rwxrwxrwx 1 root root 973176832 Jan 28 13:17 undo_1.dat

And the funny thing is that it started to create the undo_1.dat file, and then after a while said the permissions were wrong...

How to repeat:
create a logfile group with an undo file somewhere else than in the datadir.

Suggested fix:
-

Can you reproduce this or was it a one time failure?
I wasn't able to reproduce it, 

  CREATE LOGFILE GROUP lg_1 ADD UNDOFILE '/data2/users/hartmut/undo_1.dat'
   INITIAL_SIZE=4096M UNDO_BUFFER_SIZE=128M ENGINE=ndb; 

worked just fine for me on our support cluster test machines ...

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".