Last Updated: 2015-06-01
Condition or Error
A backup was taken with xtrabackup-full from the master. Slave appears to not be configured for xtrabackup-full, which results in there being issues with the restore. How can we configure the slave to use xtrabackup-full for restore?
Causes
The underlying cause and indication is that the xtrabackup has not been installed properly on the slave, or not installed at all, at the point when Tungsten Clustering was being installed. The following will be seen in the status output after a failed restore:
minimumStoredSeqNo : -1 offlineRequests : NONE pendingError : Unable to spawn restore request pendingErrorCode : NONE pendingErrorEventId : NONE pendingErrorSeqno : -1 pendingExceptionMessage: Backup agent name not found: xtrabackup-full
Probably the other hosts didn't require this setting specifically because xtrabackup was installed and detected when Tungsten Clustering was installed on them.
Rectifications
The steps you need are:
Install xtrabackup if not already installed on the slave in question
If using the INI configuration method, add the line below
to /etc/tungsten/tungsten.ini:
backup-method=xtrabackup-full
And then run tpm update on that slave host to update the configuration.
Or if using staging method, update the configuration using:
shell> tpm update --backup-method=xtrabackup-full