On 07/17/2018 10:21 AM, Jehan-Guillaume (ioguix) de Rorthais wrote:
> On Tue, 17 Jul 2018 10:11:50 +0200
> ...and check this one: https://github.com/OPMDG/check_pgactivity/
>
> It uses bloat queries for tables and btree indexes Adrien Nayrat was pointing
> earlier in this thread.
>
> In fact, both queries in check_pgactivity were written because the bloat check
> in check_postgres was considering **all** fields were in **all***
> indexes..
not accurately, since it is excluding a few things.
from the docs:
'Tables must have at least 10 pages, and indexes at least 15, before
they can be considered by this test.'
+ you can include and exclude objects based on your taste, same as in
check_pgactivity.
The only 'drawback' of check_postgres.pl is that it checks indexes and
tables's bloat in one go. (but: if your object's names are normalized,
it should not be difficult to include or exclude them)
I do not consider it a drawback, but you are free to pick your poison...
.Which is quite a large approximation...I don't know if this is still
> the case though.
While i think both tools might fit Alessandro's purpose, please note
that check_pgactivity is **only** checking for btree indexes (which are
the default ones, and the proven-to-get-bloated-quickly)
If I were you (both), I would monitor **all** indexes (and yes! tables
too), since one day you might realize it was actually a good idea to do so.
regards,
fabio pardi