MySQL Bugs: #37875: ERROR 1296 (HY000) at line 1151: Got error 902 'Out of memory in Ndb Kernel, ord

Bug #37875	ERROR 1296 (HY000) at line 1151: Got error 902 'Out of memory in Ndb Kernel, ord
Submitted:	4 Jul 2008 14:05	Modified:	14 Oct 2008 8:26
Reporter:	Shueb Khan	Email Updates:
Status:	Not a Bug	Impact on me:	None
Category:	MySQL Cluster: Cluster (NDB) storage engine	Severity:	S2 (Serious)
Version:	5.0.41	OS:	Linux
Assigned to:		CPU Architecture:	Any

Description:
We used the "ndb_size.pl" script to retrieve the amount of DataMemory used by each of our databases, and the total (around 600) is well below the limit that we set (1800). We guess that this script does not accurately reports the DataMemory amount. Is there another mean to know the real amount used by our DBs?

root@localhost on 5.0.41>INSERT INTO `questionnaire` (`id`, `member_id`, `anciennete`, `representation`, `but_achat`, `origine_achat`, `grande_maison`, `grande_surface`, `optin`, `created_on`) VALUES ('34290','22151','1','1','1|2|3|0','2','3','1',NULLIF('', ''),'2008-06-24'); 

ERROR 1296 (HY000): Got error 902 'Out of memory in Ndb Kernel, ordered index data (increase DataMemory)' from ndbcluster 

How to repeat:
It can always be repeated, since we have the error each time we try to insert some data in a table

config.ini

Attachment: config.ini (application/octet-stream, text), 2.59 KiB.

ndb logs-1

Attachment: ndb_logs-1.zip (application/x-zip-compressed, text), 277.12 KiB.

ndb logs-2

Attachment: ndb_logs-2.zip (application/x-zip-compressed, text), 244.46 KiB.

ndb traces-1

Attachment: ndb_traces-1.zip (application/x-zip-compressed, text), 195.93 KiB.

ndb traces-2

Attachment: ndb_traces-2.zip (application/x-zip-compressed, text), 207.84 KiB.

ndb traces-3

Attachment: ndb_traces-3.zip (application/x-zip-compressed, text), 240.07 KiB.

ndb traces-4

Attachment: ndb_traces-4.zip (application/x-zip-compressed, text), 139.95 KiB.

I received a mail which informed me that this has been updated, but I couldnt see any updates here.
Could you please tell me how I can check new updates on my problem.

Thanks

Thank you for the report.

I changed category to more appropriate: Server:DML -> Server:Cluster

The perl script only reports an estimate based on rather educated guesses, it is not guaranteed to produce exact results.

A 3:1 difference is strange though, but without knowing the exact data you are trying to convert it is next to impossible to know why the script doesn't produce more accurate results ...

Thanks Hartmut

So is this also related to the error we get while inserting data?
Could you suggest how we can resolve this error

ERROR 1296 (HY000): Got error 902 'Out of memory in Ndb Kernel, ordered index data
(increase DataMemory)' from ndbcluster

As the error message says: increase DataMemory?

But this error message comes even when an attempt to insert a single row is made

Could restarting the cluster solve this problem ?
How much time could it take if we plan to restart it.. it has 2 nodes.
And I have been informed that restarting this database has failed previously due to some errors, and required recovery. So I am a bit reluctant whether it is advised to restart it, and what procedure should I follow to restart it.

Even inserting a single row might make a memory allocation step necessary if the already allocated memory is full. A rolling --initial restart might help in reorganizing memory a bit but is too risky if you are already out of memory.

You have to increase DataMemory as suggested by the error message and do a rolling restart then.