Re: missing toast table for pg_policy - Mailing list pgsql-hackers

From John Naylor
Subject Re: missing toast table for pg_policy
Date
Msg-id CAJVSVGVfVuatPOX6Rzz6zKbTGsozC=K+L+W=q1=rLPUM3=JoBw@mail.gmail.com
Whole thread Raw
In response to Re: missing toast table for pg_policy  (Michael Paquier <michael@paquier.xyz>)
List pgsql-hackers
On 2/20/18, Michael Paquier <michael@paquier.xyz> wrote:
> Regression tests of pg_upgrade are failing as follows:
> New cluster database "postgres" is not empty
> Failure, exiting
> + rm -rf /tmp/pg_upgrade_check-Xn0ZLe

I looked into this briefly. The error comes from
check_new_cluster_is_empty() in src/bin/pg_upgrade/check.c, which
contains the comment

/* pg_largeobject and its index should be skipped */

If you don't create a toast table for pg_largeobject and
pg_largeobject_metadata, the pg_upgrade regression test passes.
Revised addendum attached. Adding two more exceptions to my alternate
implementation starts to make it ugly, so I haven't updated it for
now.

-John Naylor

> Michael
>

Attachment

pgsql-hackers by date:

Previous
From: Jeevan Chalke
Date:
Subject: Re: Server crashed with TRAP: FailedAssertion("!(parallel_workers >0)" when partitionwise_aggregate true.
Next
From: Jeremy Finzel
Date:
Subject: Re: found xmin from before relfrozenxid on pg_catalog.pg_authid