Date: Mon, 8 Jun 2020 13:37:07 -0400
From: Alvaro Herrera <alvherre@2ndquadrant.com>
To: Andrey Klychkov <aaklychkov@mail.ru>
Cc: Michael Paquier <michael@paquier.xyz>,
	Peter Eisentraut <peter.eisentraut@2ndquadrant.com>,
	pgsql-bugs@lists.postgresql.org, Tom Lane <tgl@sss.pgh.pa.us>
Subject: Re:
Message-ID: <20200608173707.GA18596@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <1590581418.481387288@f493.i.mail.ru>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On 2020-May-27, Andrey Klychkov wrote:

> =======
> BEFORE:
> =======
>
> Indexes:
>     "act_re_procdef_pkey" PRIMARY KEY, btree (id_)
>     "act_uniq_procdef" UNIQUE CONSTRAINT, btree (key_, version_, tenant_id_)
>     "act_re_procdef_deployment_id_idx" btree (deployment_id_)

> =====
> AFTER
> =====
>
> Indexes:
>     "act_re_procdef_pkey" PRIMARY KEY, btree (id_)
>     "act_re_procdef_pkey_ccnew" UNIQUE, btree (id_) INVALID
>     "act_uniq_procdef" UNIQUE CONSTRAINT, btree (key_, version_, tenant_id_)
>     "act_uniq_procdef_ccnew" UNIQUE, btree (key_, version_, tenant_id_) INVALID
>     "act_re_procdef_deployment_id_idx" btree (deployment_id_)
>     "act_re_procdef_deployment_id_idx_ccnew" btree (deployment_id_) INVALID

> I got the error related to the one UNIQUE index, why all the new indexes are invalid?

> >There is a paragraph about the handling of invalid indexes on the
> >reindex page:
> >https://www.postgresql.org/docs/devel/sql-reindex.html#SQL-REINDEX-CONCURRENTLY
> >"The recommended recovery method in such cases is to drop the invalid
> >index and try again to perform REINDEX CONCURRENTLY.

I think this part remains unanswered: why are there *three* invalid
indexes, and not two?  It makes sense for the UNIQUE indexes to acquire
invalid duplicates, but strangely we also have a
act_re_procdef_deployment_id_idx_ccnew which is *not* unique.  What
happened there?

(I also wonder if it wouldn't make more sense to reindexdb to attempt to
drop indexes that it couldn't complete.)

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services