Re: cache lookup failed for function 1 on fmgr_info_cxt_security,fmgr.c:184

I did not re-initdb, just use the old pgdata. I also did not recreate 
current database.
Some of the database has value 1 for lanvalidator, some has 2247.

I'll check with the team how the lanvalidator got 1 for the value. I 
will report back here.

Thanks,
Soni.

On 23/04/2019 10:51, Tom Lane wrote:
> Soni <diptatapa@gmail.com> writes:
>> Here it is:
>> select * from pg_language;
>>    lanname  | lanowner | lanispl | lanpltrusted | lanplcallfoid |
>> laninline | lanvalidator | lanacl
>> ----------+----------+---------+--------------+---------------+-----------+--------------+--------
>>    internal |       10 | f       | f            |             0 |
>> 0 |         2246 |
>>    sql      |       10 | f       | t            |             0 |
>> 0 |         2248 |
>>    plpgsql  |       10 | t       | t            |         13793 |
>> 13794 |        13795 |
>>    c        |       10 | f       | f            |             0 |
>> 0 |            1 |
>> (4 rows)
> Hm.  That doesn't look like data corruption: that looks like
> somebody actually did
>
> update pg_language set lanvalidator = 1 where lanname = 'c';
>
> or something equivalent to that.  Needless to say, that was
> not a good idea.
>
> You said you reinstalled the postgres executables, but did you
> re-initdb?  Or even just re-create the current database?
> I'm wondering if it's like that in every database of your cluster.
>
> It'd be possible to fix this particular database by setting
> lanvalidator back to its correct value (2247) but it'd be
> wise to first inquire as to how it got like this.
>
>             regards, tom lane