Re: strange behavior of .pgpass file

On 6/20/23 11:59, Atul Kumar wrote:
> Hi,
>
> I found some strange behaviour of postgres superuser password in my
> existing cluster, below is the basic outline of the setup.
> 1. version - postgres 12
> 2. replication - streaming replication async
> 3. OS- centos7
> 4. One Master, One Standby
>
> I have identical pgpass files on both server postgres home directory.
>
> So when I execute below command on slave node:
> psql -h slave_hostname/ip -U postgres -d postgres -p 5432 -c "select
> pg_is_in_recovery"
>
> I get error of password authentication:
>
> psql: error: FATAL:  password authentication failed for user "postgres"
>
> password retrieved from file "/homedirectorypath/.pgpass"

I'm going to say this is failing because per:

https://www.postgresql.org/docs/15/libpq-pgpass.html

hostname:port:database:username:password

and when you are running it the hostname is not matching what you think
it is and the wrong password is being returned. Whereas the example
below is matching correctly.

>
>
> But when I run the same command on master node:
>
> psql -h slave_hostname/ip -U postgres -d postgres -p 5432 -c "select
> pg_is_in_recovery"
>
>
> I don't get any errors and get the expected output as "t".
>
>
> *_Note: the passwords in and path of both pgpass files are identical._*
>
>
> then why am I getting errors on the slave node for the same command even
> after having everything the same ?
>
>
> Also, I tried the -W to enforce the password of postgres user but got
> the same issue on slave and no issue on master although the password is
> the same.
>
>
> What am I missing here ? Please suggest.
>
>
>
>
> Regards,
> Atul
>
>
>
>

--
Adrian Klaver
adrian(dot)klaver(at)aklaver(dot)com