MySQL Bugs: #70908: MEM3 shows what seem to be some RBR statements via replication, not all

Bug #70908	MEM3 shows what seem to be some RBR statements via replication, not all
Submitted:	14 Nov 2013 13:37	Modified:	11 Dec 2013 15:48
Reporter:	Simon Mudd (OCA)	Email Updates:
Status:	Not a Bug	Impact on me:	None
Category:	MySQL Enterprise Monitor: Server	Severity:	S3 (Non-critical)
Version:	3.0.3	OS:	Any
Assigned to:		CPU Architecture:	Any

Description:
I see on a recently installed MEM server on a busy slave with no clients that there are a lot of BEGIN statements showing in the BrowseQueries.action page.

Yet I see no other queries. This information seems to be coming in from replication.
This seems to not be consistent as I'd expect all data from the SQL thread to be collected, or none.

The server is running 5.6.14.

How to repeat:
setup a replication chain master/slave using RBR and writing to a different database on the master, with BEGIN ; INSERT ...; COMMIT; and then see if you see something appear in MEM on the slave.

Suggested fix:
1. Bug: Be consistent and show nothing or something.

2. Feature request: Ideally show all statements executed by the SQL thread if using SBR.

3. Feature request: If using you can not provide the statements but you could provide a pseudo statement providing the statement type (INSERT/UPDATE/DELETE/... etc and table name and then the number of rows and size of the update "event" that this triggers. That information would be most useful. In fact I believe that it's possible to include the source SQL statement but given the way RBR operatives the values I mentioned would be appropriate as the actual SQL is not executed on the slave.

I'll see if I can provide a screenshot showing this as I see it.

Note: period shown is last 30 minutes.

Hello Simon,

Thank you for the bug report.
Verified as reported.

Thanks,
Umesh

This is not a bug, we do actually get all SBR statements executed by the SQL thread when using P_S, and get "as much as we can" in a consistent way to all other threads.

RBR events will have a very different "shape" to statement based events, and are not available from P_S at the moment. There's essentially "nothing to do here" from the MEM side at the moment, until that arrives.