MySQL Bugs: #69955: Got temporary error 20008 'Query aborted due to out of query memory'

Bug #69955	Got temporary error 20008 'Query aborted due to out of query memory'
Submitted:	7 Aug 2013 15:37	Modified:	6 Apr 2016 15:15
Reporter:	Mateusz Kamola	Email Updates:
Status:	Can't repeat	Impact on me:	None
Category:	MySQL Cluster: Cluster (NDB) storage engine	Severity:	S2 (Serious)
Version:	7.2.12	OS:	Linux (Debian GNU/Linux wheezy)
Assigned to:	MySQL Verification Team	CPU Architecture:	Any
Tags:	out of query memory

Description:
We are running MySQL Cluster on two machines.
Both machines run one Data Node and one MySQL Node.

Since yesterday we are getting this error:
(HY000) SQLSTATE[HY000]: General error: 1297 Got temporary error 20008 'Query aborted due to out of query memory' from NDBCLUSTER

mysql_errors.log contains same errors:
130807 17:25:53 [ERROR] Got error 20008 when reading table './api/table_name'

Error occurs on both machines, several times a second with few queries.
Explain for one of the queries:
1	SIMPLE	pr	ref	PRIMARY,profileId,date_created_at	profileId	7	const,const	11	Parent of 2 pushed join@1; Using where; Using temporary; Using filesort
1	SIMPLE	ref	ref	word,probe_id	probe_id	8	api.pr.id	19	Child of 'pr' in pushed join@1
1	SIMPLE	d	ref	word	word	767	api.ref.word	97	Using where

Don't know what caused error to appear (it's happening for the first time, we are using cluster for about a year).

How to repeat:
Don't know yet.

ndb_error_report

Attachment: ndb_error_report_20130807173530.tar.bz2 (application/x-bzip2, text), 935 bytes.

Hi Mateusz,

Appears something went wrong when you created the error report since it only contains the config.ini file.

$ tar tjf ndb_error_report_20130807173530.tar.bz2 
ndb_error_report_20130807173530/
ndb_error_report_20130807173530/config.ini

But from your configuration I can see you are running the management node on same host as a data node:
-->
[NDB_MGMD]
NodeId = 1
HostName = 192.168.0.4
ArbitrationRank = 1
DataDir = /opt/mysql/server-5.5/ndb_data

[NDBD]
NodeId = 2
HostName = 192.168.0.4
-->

This is generally a bad idea since if the host would fail you would lose both the data node and management node at the same time. Assuming the management node is also arbitrator the other data node will shut down to avoid a split brain scenario.

Could you please recreate the ndb_error_report and verify it contains the files needed? Then attach it together with the MySQL error logs from both SQL nodes.

/Gustaf

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".