MySQL Bugs: #44705: Memory usage statistics events are flooding the cluster log (80%/90%/100%))

Bug #44705	Memory usage statistics events are flooding the cluster log (80%/90%/100%))
Submitted:	6 May 2009 21:40	Modified:	8 Feb 2024 13:07
Reporter:	Alexander Yu	Email Updates:
Status:	Won't fix	Impact on me:	None
Category:	MySQL Cluster: Cluster (NDB) storage engine	Severity:	S3 (Non-critical)
Version:	mysql-5.1-telco-7.0	OS:	Any
Assigned to:		CPU Architecture:	Any
Tags:	cluster log, MySQL, MySQL Cluster

Description:
Statistics events for memory usage when data/index memory reaches 80/90 percent is flooding the cluster log (Data and index memory usage (80%, 90%, and 100%) ). 

Currently this event's prio is set too low(5) which means we cannot turn it off if we just want transaction statistics info(9) for example. In some cases you're fine with running 80% of capacity.

Also, the mem usage is written every second, every N seconds would be enough.

--
2009-05-06 23:18:56 [MgmSrvr] INFO     -- Node 3: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:18:56 [MgmSrvr] INFO     -- Node 4: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:18:57 [MgmSrvr] INFO     -- Node 3: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:18:57 [MgmSrvr] INFO     -- Node 4: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:18:58 [MgmSrvr] INFO     -- Node 3: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:18:58 [MgmSrvr] INFO     -- Node 4: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:18:59 [MgmSrvr] INFO     -- Node 3: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:18:59 [MgmSrvr] INFO     -- Node 4: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:00 [MgmSrvr] INFO     -- Node 4: Operations=0
2009-05-06 23:19:00 [MgmSrvr] INFO     -- Node 3: Operations=0
2009-05-06 23:19:00 [MgmSrvr] INFO     -- Node 3: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:00 [MgmSrvr] INFO     -- Node 3: Trans. Count = 0, Commit Count = 0, Read Count = 0, Simple Read Count = 0, Write Count = 0, AttrInfo Count = 0, Concurrent Operations = 0, Abort Count = 0 Scans = 0 Range scans = 0
2009-05-06 23:19:00 [MgmSrvr] INFO     -- Node 4: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:00 [MgmSrvr] INFO     -- Node 4: Trans. Count = 0, Commit Count = 0, Read Count = 0, Simple Read Count = 0, Write Count = 0, AttrInfo Count = 0, Concurrent Operations = 0, Abort Count = 0 Scans = 0 Range scans = 0
2009-05-06 23:19:01 [MgmSrvr] INFO     -- Node 3: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:01 [MgmSrvr] INFO     -- Node 4: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:02 [MgmSrvr] INFO     -- Node 3: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:02 [MgmSrvr] INFO     -- Node 4: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:03 [MgmSrvr] INFO     -- Node 3: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:03 [MgmSrvr] INFO     -- Node 4: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:04 [MgmSrvr] INFO     -- Node 3: Data usage increased to 80(6412 32K pages of total 8000)
2009-05-06 23:19:04 [MgmSrvr] INFO     -- Node 4: Data usage increased to 80(6412 32K pages of total 8000)

How to repeat:
Use more than 80% of available data/index mem.
Enable statistics events for the cluster log.
ndb_mgm> all clusterlog statistics=8

Suggested fix:
Change the prio level or only send the mem usage event every N seconds instead of every second which it does today.

Still very annoying in MySQL cluster 7.0.9.

Workaround: don't use ndb_mgm as shell, but use -e option.

Permanent solution suggestion:
* print at start of ndb_mgm
* print the message once after each command in ndb_mgm
* stop flooding logs, but only report when there is a change, e.g. 85% -> 86% = print message in logs

Don't see how this can be a feature request. It's very annoying: flooding logs and makes ndb_mgm unusable.

Also, bit inconsistent: with MySQL Cluster 6.3.28b on Linux, logs and ndb_mgm are not flooded.

This is Log informance,so you can turn on it..

please see http://dev.mysql.com/doc/refman/5.1/en/mysql-cluster-logging-management-commands.html

Posted by developer:
 
No longer a problem.

Closing as rejected.