On 06/05/17 18:16, Tom Lane wrote:
> Petr Jelinek <petr.jelinek@2ndquadrant.com> writes:
>> On 06/05/17 17:25, Tom Lane wrote:
>>> OK, can you suggest better wording?
>
>> Something like the attached (it requires some polishing of English
>> probably).
>
> Hmm, I'm hoping for something more user-oriented. Is the corruption
> time-limited? What's an "exported snapshot" anyway, is it the same
> thing as pg_export_snapshot(), and if so what's that got to do with
> logical replication?
>
Well user does not care unless they use something like pglogical, or
bottledwater, or wal2json, etc.
Okay to explain what's happening. When you create logical replication
slot via walsender, it exports snapshot (like the one exported by
pg_export_snapshot() indeed) which corresponds to exact point in time
where the decoding will start for the slot. You can import this snapshot
to another backend and use it to copy any existing data before starting
the replication. The bugs cause that these snapshots would be corrupted
and you'd have inconsistent data (some tuples missing). One of them
would export snapshot that's only valid for system catalogs but not user
tables (the ondisk snapshot, this needs at least one preexisting slot).
The other would not guarantee that tuples needed by the snapshot weren't
removed by vacuum of HOT pruning (the window being only the time it took
to generate the snapshot).
-- Petr Jelinek http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training &
Services
