On 6/2/20 1:30 PM, Stephen Frost wrote:
Greetings,
* Ron (ronljohnso...@gmail.com) wrote:
On 6/2/20 4:59 AM, Grigory Smolkin wrote:
On 6/2/20 11:22 AM, Ron wrote:
The inability to do a point-in-time restoration of a *single* database
in a multi-db cluster is a serious -- and fundamental -- missing feature
(never to be implemented because of the fundamental design).
It is possible via 3rd party tools like pg_probackup and pgbackrest.
pgbackrest does *not* support PITR recovery of individual databases into
*new* database names in the same cluster (so that the end user can have both
the current database and an old version at the same time).
No, nothing does as PG doesn't support it as we have one WAL stream for
the entire cluster.
Right. Making WAL files specific to a database should be high on the list
of priorities.
Generally speaking, I discourage having lots of databases under one PG
cluster for exactly these kinds of reasons.
It's just two... :)
PG's individual clusters are relatively lightweight, after all.
But require a new port, and Enterprises have Processes that must be followed.
--
Angular momentum makes the world go 'round.
