MySQL Bugs: #27292: restarting a data node makes sql nodes log event buffer status every few seconds

Bug #27292	restarting a data node makes sql nodes log event buffer status every few seconds
Submitted:	20 Mar 2007 14:46	Modified:	4 Aug 2008 10:04
Reporter:	Hartmut Holzgraefe	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Cluster: Cluster (NDB) storage engine	Severity:	S3 (Non-critical)
Version:	5.1bk	OS:	Any
Assigned to:	Tomas Ulin	CPU Architecture:	Any

Description:
after restarting a data node with "id RESTART" all mysql nodes
start to write the following log messages about every 2 seconds,
so spamming the logs

2007-03-20 15:39:33 [MgmSrvr] INFO     -- Node 5: Event buffer status: used=128B(100%) alloc=128B(0%) max=0B apply_gci=15 latest_gci=31
2007-03-20 15:39:33 [MgmSrvr] INFO     -- Node 4: Event buffer status: used=128B(100%) alloc=128B(0%) max=0B apply_gci=15 latest_gci=31

How to repeat:
Set up a fresh cluster (1 ndb_mgmd, 2 ndbd (replica=2), 2 mysqld in my case),
restart one ndbd using 

  ndb_mgm -e "2 restart",

watch the ndb_#_cluster.log (e.g. with tail -f) and see the log spam to start once the data node is back up

Looks like the bug is around line 1501 in NdbEventOperationImpl.cpp.

The workaround is to comment reportStatus() on 1515, but that doesn't solve the problem.

Upgrading to P2, behind this is actually a "small memleak"

this should only show up if bin-logging is not on

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/commits/26366

ChangeSet@1.2511, 2007-05-09 12:51:37+02:00, tomas@whalegate.ndb.mysql.com +1 -0
  Bug #27292  	restarting a data node makes sql nodes log event buffer status every few seconds

Pushed into 5.1.19-beta

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 bug fix. More information about accessing the source trees is available at

    http://dev.mysql.com/doc/en/installing-source.html

Documented bugfix in 5.1.18 changelog. Noted that it was observed only in 5.1.16 and later.

This bug has to be documented more clearly, something like:

A SQL Node acting as a Slave will gradually consume more memory, each time with 4Mb. This is because some events were not consumed. It is also irrelevant whether binary logging is enabled or not and has nothing to do with data node restarting. Simply doing replication is enough to show the problem.

Fix was actually documented in the 5.1.19 changelog, not 5.1.18 (typo in my previous comment).