Repair of cluster configuration information when no CCR backup
is available
The following procedures describe how to repair missing or corrupted cluster configuration information, when no intact CCR can be found on the quorum nodes and no CCR backup is available from which the broken cluster can be recovered.
These procedures do not guarantee to recover the most recent state of all the configuration files in the CCR. Instead, they bring the CCR back into a consistent state with the most recent available version of each configuration file.
All of the procedures include the following major steps:
- Diagnose if the CCR is broken on a majority of the quorum nodes.
- Evaluate the CCR's most recent Paxos state file.
- Patch the CCR's Paxos state file.
- Verify that the CCR state is intact after patching the Paxos state file and copying back to the CCR directory.
In most cases the mmsdrrestore
--ccr-repair command restores missing or corrupted CCR files as well as or
better than the manual procedures.
- Repair of cluster configuration information when no CCR backup information is available: mmsdrrestore command
When no CCR backup information is available, you can repair missing or corrupted CCR files with the mmsdrrestore command and the --ccr-repair parameter. - Manual repair procedure for a broken single-node cluster
The CCR becomes unavailable on a quorum node due to the presence of corrupted files in the CCR committed directory. The presence of corrupted files in the CCR committed directory might be the result of a node crash or a hard power-off. - Manual repair procedure for a broken multi-node cluster
For a multi-node cluster containing more than one quorum node, the recovery procedure is similar to a single node cluster with just one quorum node. However the number of possible CCR states is different, as the quorum nodes can have different CCR states.
Parent topic: Recovery procedures