Horton

Re: Crash in BRIN minmax-multi indexes

Tomas Vondra <tomas.vondra@enterprisedb.com>

View thread View on postgr.es
From: Tomas Vondra <tomas.vondra@enterprisedb.com>
To: Zhihong Yu <zyu@yugabyte.com>, Jaime Casanova <jcasanov@systemguards.com.ec>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2021-03-31T18:27:39Z
Lists: pgsql-hackers 
On 3/31/21 8:20 PM, Zhihong Yu wrote:
> Hi,
> In build_distances():
> 
>         a1 = eranges[i].maxval;
>         a2 = eranges[i + 1].minval;
> 
> It seems there was overlap between the successive ranges, leading to
> delta = -6785000000
> 

I've been unable to reproduce this, so far :-( How exactly did you
manage to reproduce it?


The thing is - how could there be an overlap? The way we build expanded
ranges that should not be possible, I think. Can you print the ranges at
the end of fill_expanded_ranges? That should shed some light on this.


FWIW I suspect those asserts on delta may be a bit problematic due to
rounding errors. And I found one issue in the inet distance function,
because apparently

test=# select '10.2.14.243/24'::inet < '10.2.14.231/24'::inet;
 ?column?
----------
 f
(1 row)

but the delta formula currently ignores the mask. But that's a separate
issue.


regards

-- 
Tomas Vondra
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Commits

  1. Fix assert in BRIN build_distances

  2. Fix bug in brin_minmax_multi_union

  3. Fix order of parameters in BRIN minmax-multi calls

  4. Fix BRIN minmax-multi distance for inet type

  5. Fix BRIN minmax-multi distance for timetz type

  6. Fix BRIN minmax-multi distance for interval type