Re: sequences vs. synchronous replication

On 12/27/21 21:24, Peter Eisentraut wrote:
> On 24.12.21 09:04, Kyotaro Horiguchi wrote:
>> Still, as Fujii-san concerns, I'm afraid that some people may suffer
>> the degradation the patch causes.  I wonder it is acceptable to get
>> back the previous behavior by exposing SEQ_LOG_VALS itself or a
>> boolean to do that, as a 'not-recommended-to-use' variable.
>
> There is also the possibility of unlogged sequences if you want to avoid
> the WAL logging and get higher performance.

But unlogged sequences are not supported:

test=# create unlogged sequence s;
ERROR: unlogged sequences are not supported

And even if we did, what would be the behavior after crash? For tables
we discard the contents, so for sequences we'd probably discard it too
and start from scratch? That doesn't seem particularly useful.

We could also write / fsync the sequence buffer, but that has other
downsides. But that's not implemented either, and it's certainly out of
scope for this patch.

regards

--
Tomas Vondra
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company