I'd like to confirm a cluster upgrade rollback procedure.
We have a 3 node 9.x cluster - node 1 has all the databases & forests , nodes 2 & 3 are e-nodes with no databases, apart from the system databases. It also has a 4.1.1 Data Hub.
Can I confirm how to roll back the cluster to its previous 9.x state , in event of a failed upgrade , borrowing some of the instructions from "Cloning an instance" from URL below : https://help.marklogic.com/Knowledgebase/List/Index/19/administration
Before upgrade to 10.x :
- Back up the /var/opt/MarkLogic directory on each of the 3 Linux cluster nodes.
- Backup all user databases and system databases on each node ( shown below )
- Documents
- Extensions
- Fab
- Last-Login
- Meters
- Modules
- Schemas
- Security
- Triggers
Rolling back the cluster to 9.x
- De-install the failed MarkLogic 10.x upgrade on each node, starting with cluster boot node.
- Re-install MarkLogic server 9.x on each node , starting with cluster boot node.
- Restored the /var/opt/Marklogic directory ( overwriting the new directory installed by new 9.x install ) on each cluster node, starting with boot node.
- Restore the Security and Modules databases on boot node ( node 1 ).