MySQL Bugs: #45150: LOST_EVENTS not documented

Bug #45150	LOST_EVENTS not documented
Submitted:	28 May 2009 8:09	Modified:	20 Aug 2009 7:43
Reporter:	Johan Andersson	Email Updates:
Status:	Not a Bug	Impact on me:	None
Category:	MySQL Server: Replication	Severity:	S3 (Non-critical)
Version:	all (?)	OS:	Any
Assigned to:		CPU Architecture:	Any
Tags:	internals, LOST_EVENTS, replication

Description:
LOST_EVENTS not documented

How to repeat:
Buy reading the manual

Suggested fix:
Document it

Wonderful! So what are these LOST_EVENTS and where do I find out about them? :)

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".

I stumbled over this bug report after reading Johan's blog post here: http://johanandersson.blogspot.com/2009/05/ha-mysql-write-scaling-using-cluster-to.html

Johan, where can we get more information on LOST_EVENTS?

Also, is this exclusive to Cluster, or does this apply to replication generally?

i'd say he's actually talking about the GAP event,

Hi!

After reading Johan's blog entry and doing some digging around, I've reached the conclusion that this would best be documented in the Internals Manual at http://forge.mysql.com/wiki/MySQL_Internals_Binary_Log, which is not maintained by Docs, but rather by the development teams. In this case, it appears that this should be done by someone from the Replication & Backup Team.

Please assign a developer to update this material. I've set myself as a reviewer and will be glad to assist with edits once the text is available.

Thanks!

From the Internals page in question:

"Incident number codes are listed in rpl_constant.h. The only code currently used is INCIDENT_LOST_EVENTS, which indicates that there may be lost events (a "gap") in the replication stream that requires databases to be resynchronized."

Given that the event is already documented, I close this bug as a "not a bug" (the event was already documented). Re-open the bug (and tell me) if you disagree with the decision.