GA Release v1.4.500
Recovery:
- Clusters dashboard indicate per chain whether it has auto failover for master/intermediate master
- Audit-detection page provides with extra info on the crash analysis.
- This includes the analysis changelog for the failed instance
- Also suggests the related recovery (if taken) for this detection
- Audit-recovery page provides with extra info on recovery; also suggests the related discovery event
- Recognizing crashed recoveries (recoveries started by an orchestrator that crashed halfway through)
- Automatically acknowledging such crashes
- This may potentially cause an endless rolling recoveries sequence, in the case where an internal bug causes orchestrator to consistently crash on a given scenario.
- Manually invoked recoveries always override any blocks
- ApplyMySQLPromotionAfterMasterFailover config will set promoted master as writable and issue a RESET SLAVE.
- Better detection of GTID-based recovery scenario
Visibility:
- Any running orchestrator (HTTP or CLI) identifies itself in the node_health table
- including the type of invocation and the primary command it was executing
- This allows for quick understanding of what's running, when and why
- node_health_history table keeps record (for a few days) of all past orchestrator invocations, from anywhere.
Other:
Added "--skip-unresolve" command line flag