Larry Rosenman <ler@lerctr.org> writes:
> On 3/7/2012 12:31 PM, Alvaro Herrera wrote:
>> Also, this isn't limited to CLUSTER; anything that rewrites the
>> table and indexes would benefit. Meaning ALTER TABLE that does a
>> full rewrite, and also VACUUM FULL.
> +1. I think we should update ps_status as well as pg_stat_activity as
> all of these table rewrite processes are running.
-1. Updating ps_status is expensive, seriously so on some platforms.
We could likely get away with tracking progress in pg_stat_activity
though.
However, this just reminds me that tracking intrastatement progress in
pg_stat_activity has been discussed before and not much has happened.
Let's please not have a quick kluge that just addresses CLUSTER and not
any of the other aspects of that problem.
regards, tom lane