Hi all,
I hope I'm posting this in the correct place.
We use pgbouncer, as I'm sure many other people do, but we are becoming
increasingly worried by the lack of a new release since February 2016 and a
slowdown in development activity on the master branch.
https://github.com/pgbouncer/pgbouncer
> Unfortunately, that'll require locking each table and scanning it to make
> sure that the CHECK constraint isn't violated.
Actually, CHECK constraints can be added with the NOT VALID clause.
New tuples will be checked immediately, while the validation of existing tuples
can be done later
> I was wondering if there was any way to break down the creation of a new
> exclusion constraint into stages such that table locks most likely to affect
> performance during production hours are not taken.
>
> Something like:
>
> CREATE INDEX CONCURRENTLY new_index ON my_table USING gist
Hi,
I was wondering if there was any way to break down the creation of a new
exclusion constraint into stages such that table locks most likely to affect
performance during production hours are not taken.
Something like:
CREATE INDEX CONCURRENTLY new_index ON my_table USING gist (column1,
>>> In my production system, there are lot of read write operation performed
>>> every hour.
>>Apologies if this sounds patronising but I just wanted to check - you know
>>that indexes are updated automatically when write operations occur, right?
>Yes. You are correct.
OK - I think the folks
> In my production system, there are lot of read write operation performed
> every hour.
Apologies if this sounds patronising but I just wanted to check - you know that
indexes are updated automatically when write operations occur, right?
This email is confidential. If you are not the
specified at cluster
creation time using ‘initdb -X ’, which (AFAICT) just creates the
symlink for you.
Steve.
Dr. Steven Winfield
Scientist
D: +44 (0)1223 755 776
[Cantab email sig]
Is there any way to find when a table was last successfully CLUSTER'd, say from
pg_catalog?
Last analyzed and last vacuumed times are available, but I can't seem to find
anything to do with reclustering.
Thanks,
Steve.