Bug #61080 Topology not detected correctly

• Submitted: 6 May 2011 14:08
• Modified: 10 May 2011 9:21
• Reporter: Daniël van Eeden
• Status: Won't fix
• Category: MySQL Enterprise Monitor: Web
• Severity: S2 (Serious)
• Version: 2.3.3
• OS: Any
• CPU Architecture: Any

[6 May 2011 14:08] Daniël van Eeden

Description:
1. Setup a master1 and a slave1
2. Stop replication op slave1
3. Add a new slave2 with as master slave1
4. Issue 'RESET SLAVE' op slave1 (as it's now the master)

MEM will detect this
master->slave1->slave2 (master->slave1 = broken, slave1->slave2 = ok).

Before 5.5 this could be fixed by issueing a CHANGE MASTER TO MASTER_HOST='', but now it's inteded behaviour (See Bug #53989).

So MEM should detect somehow that master->slave1 is not used anymore...

This procedure is used when a master fails.

How to repeat:
See description.

Suggested fix:
Unknown.

[10 May 2011 9:21] Valeriy Kravchuk

It would be hard to fix this in frames of MEM only, without some way (provided by server) to discover that server is no longer a slave. We have internal bug report against 5.5.8+ asking about this, so both MEM and replication teams are aware of this problem.

[10 May 2011 10:34] Mark Leith

The work around at the moment is to issue RESET SLAVE, and then restart the instance.

[13 Mar 2014 13:37] Omer Barnir

This bug is not scheduled to be fixed at this time.