Re: strange behavior (corruption?) of large production

Tom Lane wrote:
> Joe Conway <mail(at)joeconway(dot)com> writes:
>
>>We have very strange behavior from an internal production database.
>>There are multiple symptoms, all pointing to a problem with clusterwide
>>tables. For example:
>
>
> "psql -l" really should produce the same results as doing "\l" in the
> template1 database. Does it?

Sorry -- on my last post "psql -l" was pointing to the wrong place.

postgres(at)csdfds1:~> psql -p 5433 -l
List of databases
Name | Owner | Encoding
------+-------+----------
(0 rows)

postgres(at)csdfds1:~> psql -U postgres -p 5433 template1
Welcome to psql 7.4.8, the PostgreSQL interactive terminal.

Type: \copyright for distribution terms
\h for help with SQL commands
\? for help on internal slash commands
\g or terminate with semicolon to execute query
\q to quit

template1=# \l
List of databases
Name | Owner | Encoding
------+-------+----------
(0 rows)

So they agree in template1 and cyspec databases.

> If so, the next thing to look at is probably whether the pg_class and pg_attribute entries for pg_database
> look the same in template1 and in cyspec. Similarly you could compare
> what pg_shadow looks like from different databases, and what the local
> system tables have as entries for it.

Did that - they look the same. We did a simple
cat 1262 | less
to view the contents of pg_database and indeed the data looks to be
there. Similarly a simple cat of the pg_shadow relfilenode shows the
colprod user. Is it possible that we have corrupted shared memory, and a
database restart will fix the problem? We didn't want to restart until
the forensics were done.

Joe