Re: Index corruption with CREATE INDEX CONCURRENTLY

El 05/02/17 a las 21:57, Tomas Vondra escribió:
>
> +1 to not rushing fixes into releases. While I think we now finally
> understand the mechanics of this bug, the fact that we came up with
> three different fixes in this thread, only to discover issues with each
> of them, warrants some caution.

I agree also with Robert on not rushing the patch. My point was if we
had to rush the release.

> OTOH I disagree with the notion that bugs that are not driven by user
> reports are somehow less severe. Some data corruption bugs cause quite
> visible breakage - segfaults, immediate crashes, etc. Those are pretty
> clear bugs, and are reported by users.
>
> Other data corruption bugs are much more subtle - for example this bug
> may lead to incorrect results to some queries, failing to detect values
> violating UNIQUE constraints, issues with foreign keys, etc.

I was recalling just yesterday after sending the mail a logical
replication setup we did on a 9.3 server of a customer which brought up
data inconsistencies on the primary key of one of the tables. The table
had duplicate values.

As Tomas says, it's subtle and hard to find unless you constantly run
index checks (query a sample of the data from the heap and from the
index and check they match). In our case, the customer was not aware of
the dups until we found them.

Regards,

--
Martín Marqués http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services