Dhaval Shah wrote:
> The optimization step I am looking for is as follows, instead of doing
> a tar/gzip of the data directory, I am planning to do a pg_dumpall of
> the primary. That is, the new steps will be:
>
> 0. Ensure that no application connects to the db.
> 1. On the primary, do a pg_dumpall
> That way, I figure I do not have to move the entire data directory
> around. I just need the base schema and data contents. The question
> is, will the above setup work? Any gotchas?
Won't work - the WAL refers to disk blocks, not primary keys etc. You
need the actual disk files.
--
Richard Huxton
Archonet Ltd