I'm wondering about the validity of using the pg_checksums utility to enable checksum's on a streaming replication standby, and then promoting that standby as a way to enable checksums on existing clusters.
I've tested the process out, and it "works" (by works I mean doesn't blow up or log any errors). But this seems far enough outside of supported territory that I'm curious what others think.
The process is:
- Primary with checksums off
- create a streaming replica off that primary
- stop the secondary
- enable checksums on the secondary with the pg_checksums utility
- start the replica
- promote the replica
I've thrown load at it while the it was replicating from checksums off -> checksums on, promoted it, and verified the checksums after with pg_checksums.
Any thoughts on this approach?
Thanks,
Brad