Idea ID: 2834585

DMA - SQL Cluster patching with roll back in the same flow

Status : Waiting for Votes
7 months ago

Scenario

The customer asks for a new development of the SQL Cluster DMA workflow that should roll back the patch on the whole cluster if at least on one node of the cluster the patch fails.
The ootb WF patches all the nodes of the cluster and just terminates in error if one patching subflow fails.
The need is to roll back the patched nodes in the same wf and terminate the mail flow leaving the whole cluster at the same patch level (all nodes patched or all nodes not patched).

Workaround

run the roll back for SQL standalone

Unacceptable

1. if the patching fails on some nodes of the cluster the wf leaves the cluster in an inconsistent state (it's not patched the whole cluster)
2. using the workaround the customer has to run a single flow for each node since it is not supported the rolling modality for roll back (as for cluster SQL patching)

Required

the flow should leave the cluster in a consistent state:
all nodes patched
or
all nodes NOT patched (rolling back the ones terminated in error)