Horton

Re: REINDEX INDEX results in a crash for an index of pg_class since 9.6

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Michael Paquier <michael@paquier.xyz>, Postgres hackers <pgsql-hackers@lists.postgresql.org>
Date: 2019-04-30T03:03:42Z
Lists: pgsql-hackers 
Hi,

On 2019-04-29 15:09:24 -0700, Andres Freund wrote:
> On 2019-04-29 18:07:07 -0400, Tom Lane wrote:
> > I wrote:
> > > Andres Freund <andres@anarazel.de> writes:
> > >> Taking this as a WIP, what do you think?
> > 
> > > Seems generally about right.
> > 
> > Andres, are you pushing this forward?  Next week's minor releases
> > are coming up fast, and we're going to need to adapt the HEAD patch
> > significantly for the back branches AFAICS.  So there's little time
> > to spare.
> 
> Yea. I'm testing the backbranch'd bits (much simpler) and writing the
> commit message atm.

I've pushed the master bits, and the other branches are running
check-world right now and I'll push soon unless something breaks (it's a
bit annoying that <= 9.6 can't run check-world in parallel...).

Turns out, I was confused, and there wasn't much pre-existing breakage
in RelationSetNewRelfilenode() (I guess I must have been thinking of
ATExecSetTableSpace?). That part was broken in d25f519107, I should have
caught this while reviewing and signficantly evolving the code in that
commit, mea culpa.

Greetings,

Andres Freund

Commits

  1. Remove reindex_catalog test from test schedules.

  2. Remove RelationSetIndexList().

  3. Fix reindexing of pg_class indexes some more.

  4. Run catalog reindexing test from 3dbb317d32 serially, to avoid deadlocks.

  5. Fix potential assertion failure when reindexing a pg_class index.

  6. Fix several recently introduced issues around handling new relation forks.

  7. Avoid VACUUM FULL altogether in initdb.

  8. Introduce replication progress tracking infrastructure.

  9. initdb: remove unnecessary VACUUM FULL