pt-slave-restart doesn't work with multi source replication in 5.7

Bug #1634944 reported by Francisco Bordenave
18
This bug affects 4 people
Affects Status Importance Assigned to Milestone
Percona Toolkit moved to https://jira.percona.com/projects/PT
Confirmed
Medium
Carlos Salguero

Bug Description

When you setup pt-slave-restart in a multi sourced slave it doesn't work due several reasons:
1- It only checks first channel reported in SHOW SLAVE STATUS output so if failing channel is the second one it will never notice.
2- In a multi sourced slave whenever you set SQL_SLAVE_SKIP_COUNTER >1 you are forced to run START SLAVE FOR CHANNEL 'channel_name' so you can't start it globally with just "START SLAVE"

Ideally we should be able to support monitoring either single or multi sourced slaves (i.e. adding a channel variable) by supporting FOR CHANNEL syntax.

Possible scenarios:
1- Single channel (or single master) slave. Syntax is the same as always:
SHOW SLAVE STATUS
START/STOP SLAVE

2- Multiple channels: accept --channel flag where we can define which channel to monitor. Syntax needs to support:
SHOW SLAVE STATUS FOR CHANNEL 'channel_name'
START/STOP SLAVE FOR CHANNEL 'channel_name'

Changed in percona-toolkit:
importance: Undecided → Medium
assignee: nobody → Carlos Salguero (carlos-salguero)
milestone: none → 2.2.20
Changed in percona-toolkit:
milestone: 2.2.20 → 2.2.21
Revision history for this message
Sveta Smirnova (svetasmirnova) wrote :

Thank you for the reasonable feature request.

Changed in percona-toolkit:
status: New → Confirmed
Revision history for this message
Shahriyar Rzayev (rzayev-sehriyar) wrote :

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PT-712

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.