I had a mysql (Percona 5.6) replication master go down, and found the following errors in the logs:
[Note] InnoDB: Database was not shutdown normally!
[Note] InnoDB: Starting crash recovery.
[Note] InnoDB: Reading tablespace information from the .ibd files...
[Note] InnoDB: Restoring possible half-written data pages
[Note] InnoDB: from the doublewrite buffer...
InnoDB: Doing recovery: scanned up to log sequence number 2138860811264
InnoDB: Doing recovery: scanned up to log sequence number 2138860844694
InnoDB: Apply batch completed
InnoDB: Last MySQL binlog file position 0 825950145, file name replica-1080905343-bin.002410
[Note] InnoDB: 128 rollback segment(s) are active.
[Note] InnoDB: Waiting for purge to start
[Note] InnoDB: 5.6.17 started; log sequence number 2138860844694
[Note] Recovering after a crash using /var/log/mysql/replica-1080905343-bin
[Note] Starting crash recovery...
[Note] Crash recovery finished.
/usr/sbin/mysqld: Malformed GTID set encoding.
/usr/sbin/mysqld: Malformed GTID set encoding.
[ERROR] Aborting
The crucial bit here seems to be the Malformed GTID set encoding, but I can find very little information about this error.
In particular, I’d like to have an idea of what could have caused it, as well as how to fix it with a minimum of data loss.