We’ve been bitten by the [Bug 27097] which cause a crash as soon as one of the node of the cluster is restarted.
Our problem is how to recover from this bug.
As soon as two nodes are unsealed, they crash and restart in unseal state.
I thought about several ways in prefered order :
- modify the name of the entries in vault database
- suppress the entries that cause the crash without loosing all the secrets stored in the database
- drop the whole prefix
- Start the vault without the engines that crashs (ldap secret engine)
Is any of those wishes doable ?
How to you recover all the data from a crashed prefix ?
Or maybe there is another way that I didn’t think of ?
Thanks for any help.