Bug #48201 New Option To Control LCP Before Shutdown
Submitted: 21 Oct 2009 9:57
Reporter: Mikiya Okuno Email Updates:
Status: Verified Impact on me:
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S4 (Feature request)
Version:mysql-5.1-telco-7.0 OS:Any
Assigned to: CPU Architecture:Any
Tags: 7.x

[21 Oct 2009 9:57] Mikiya Okuno
Currently, it is possible that ndb[mt]d may take long if there are lots of dirty redo entries in redo log files. This can be caused if LCP wasn't performed long before shutdown. So, we can avoid the problem by forcing LCP before shutdown using DUMP command like below:

shell> ndb_mgm -e 'ALL DUMP 7099'

However, this isn't useful enough. We need either of...

o A new command to do a same task, like "SHUTDOWN WITH CHECKPOINT". This could be much better intuitive and understood.
o A new option to force LCP when shutdown to allow users to reduce time taken to restart without any user intervention. The option name could be like "ForceLCPBeforeShutdown"

Perform LCP after disconnecting API nodes may allow data nodes to have no pending redo logs, and this could be the best.

How to repeat:

Suggested fix: