Re: Regarding ambulkdelete, amvacuumcleanup index methods - Mailing list pgsql-hackers

From Abinaya k
Subject Re: Regarding ambulkdelete, amvacuumcleanup index methods
Date
Msg-id CAEtC32P9hWKbx7q7jt24WCebvn3zEW0oFCnv491+H3gsc-H+bA@mail.gmail.com
Whole thread Raw
In response to Regarding ambulkdelete, amvacuumcleanup index methods  (Abinaya k <abinaya.k.496@gmail.com>)
Responses Re: Regarding ambulkdelete, amvacuumcleanup index methods
List pgsql-hackers
Thanks for your response.

Hope those stats will be used by Query Planner.

So, just for my understanding, if i don't return stats (returning NULL from index_bulk_delete and index_vacuum_cleanup functions), Query Planner will not consider my Index as part of Query Path, beyond that i don't expect any bad consequences.

Please clarify me, if i'm wrong.

Regards,
Abinaya K



On Wed, Jan 24, 2018 at 1:24 PM, Masahiko Sawada <sawada.mshk@gmail.com> wrote:
On Wed, Jan 24, 2018 at 1:27 PM, Abinaya k <abinaya.k.496@gmail.com> wrote:
> Hai all,
>       We are building In-memory index extension for postgres. We would
> capture table inserts, updates, deletes using triggers. During vacuum
> operation, postgres would give calls to ambulkdelete, amvacuumcleanup (as
> part of index cleanup). As we handle all updates, deletes using triggers, we
> don't have to do any index cleanup in ambulkdelete. But, what stats should i
> return from ambulkdelete and amvacuumcleanup? Is that necessary to return
> stats from ambulkdelete and amvacuumcleanup ?

Both ambulkdelete and amvacuumcleanup return an IndexBulkDeleteResult.
If you return a non-NULL value, the values of returned
IndexBulkDeleteResult are used for updating the index statistics and
reporting the statistics of bulk deletion in lazy_cleanup_index. For
example, num_pages and num_index_tuples are used for updating
pg_class.relpages and pg_class.reltuples. But if you return NULL from
them, these are skipped.

Regards,

--
Masahiko Sawada
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center


pgsql-hackers by date:

Previous
From: Marina Polyakova
Date:
Subject: Re: WIP Patch: Precalculate stable functions, infrastructure v1
Next
From: Pierre Ducroquet
Date:
Subject: Re: JIT compiling with LLVM v9.0