Extended replication variable capture

[Simon Kuhn]

Extend and improve the capture of replication variables in three ways:

1. Correct innobackupex to get the file and position from the last executed SQL statement rather than the last statement read by the IO thread:

Master_Log_File should be changed to Relay_Master_Log_File, because the latter represents the log file from which the last SQL statement was executed (rather than the log that is being read from).

Master_Log_Pos should be changed to Exec_Master_Log_Pos for clarity -- this is actually already the variable that is matched due to the order of the output of SHOW SLAVE STATUS, which prints:

        Read_Master_Log_Pos: 80
        [...]
        Exec_Master_Log_Pos: 79

both of which are matched by /Master_Log_Pos:\s*(\S*)\s*$/.

2. Add (perhaps optionally) specific master host connectivity variables to xtrabackup_slave_info:

We know MASTER_HOST and MASTER_USER, and can output them. This may not always be desirable, but is useful when cloning an existing slave to make another slave. My included patch doesn't wrap them in a command line flag option, but it could be added.

MASTER_PASSWORD is also added, but blank, since we can't get it.

3. Capture relay log and position

Add Relay_Log_File, Relay_Log_Pos in a comment in xtrabackup_slave_info for cross-referencing purposes.

[Simon Kuhn (simon.kuhn)]

Note that /Master_Log_File:\s*(\S*)\s*$/ will match Relay_Master_Log_File and do the right thing because of the order of SHOW SLAVE STATUS (just like Master_Log_Pos), but I still feel it should be changed for clarity.

[Alexey Kopytov (akopytov)]

Part #1 has also been reported as a separate bug #1099323. Let's aim to implement parts #2 and #3 for XB 2.1.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PXB-982