MySQL Bugs: #8918: Cluster crashes after first query

Bug #8918	Cluster crashes after first query
Submitted:	3 Mar 2005 10:01	Modified:	13 Jun 2005 16:10
Reporter:	Jörg Nowak	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Cluster: Cluster (NDB) storage engine	Severity:	S1 (Critical)
Version:	MySQL MAX 4.1.10	OS:	Solaris (Solaris 9 64 bit)
Assigned to:	Jonas Oreland	CPU Architecture:	Any

Description:
Hello,
I configured a cluster in 2 Sunfire E2900, 16G Ram, 4 Dual core processors running Solaris 9, 64bit. Each SunFire machine was allocated 2 ndb nodes and 1 mysqld.

MYSQL version i installed is mysql-max-4.1.10 64 bit version for solaris.

Initially, I gave the value for [NDB DEFAULT] as

------------------------------------------
DataMemory : 3500M
IndexMemory: 700M
------------------------------------------

and i have 6 tables in the cluster database. I populated with 3 Million Entries in each table.

Every thing was working fine. As i had 16G ram, i thought of increasing the memory for the ndb nodes and want to insert more records in the cluster database.

Then i made the following changes

------------------------------------------
DataMemory : 5G
IndexMemory: 1G
------------------------------------------

Then, restarted the cluster. The cluster started successfully.
But as soon as i made my first query from the console of mysql

------------------------------------------
mysql>select count(*) from CFG; (Where CFG is one of the 6 tables i have)
------------------------------------------

all the cluster nodes crashed and in the logs i could find the following message

The log files and trace files are attached.

How to repeat:
Set the
DataMemory to 3500M and IndexMemory to 700M.

Create more than 1 table in the mysql cluster and populate the tables with more than 1Million entries.

Now change the DataMemory to %Gnd IndexMemory to 1G.

and restart the cluster

Now after the cluster is started, query in each table from the console like

mysql> select count(*) from table-name.

You will then hit the bug.

Log file

Attachment: bug-info (text/plain), 719 bytes.

Trace file for the "pointer too large" error. (part of it because too large to attach)

Attachment: ndb_3_trace.log.6 (application/octet-stream, text), 123.10 KiB.

Hi Joerg,

Thanks for you effort.

In the error log (bug info) the first query points to ndb_3_trace.log.4
But you uploaded ndb_3_trace.log.6.
Furthermore, since you "cut" the trace file, it it almost useless.

PS.
  The second crash is a timeout on total startup time.
  I'll improve the printout
DS.

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".

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/internals/25794

pushed to 4.1.13 and 5.0.8

Thank you for your bug report. This issue has been committed to our
source repository of that product and will be incorporated into the
next release.

If necessary, you can access the source repository and build the latest
available version, including the bugfix, yourself. More information 
about accessing the source trees is available at
    http://www.mysql.com/doc/en/Installing_source_tree.html

Additional info:

Documented in Change History for version 4.1.13 and 5.0.8. Closed.