Re: Pgadmin III connection error

Olimpio wrote:
> blacknoz(at)club-internet(dot)fr wrote:
>
>
>>----Message d'origine----
>>
>>
>>
>>>Date: Thu, 08 Sep 2005 15:18:20 -0300
>>>De: Olimpio <olimpio(at)uaivip(dot)com(dot)br>
>>>A: blacknoz(at)club-internet(dot)fr
>>>Sujet: Pgadmin III connection error
>>>
>>>Raphael,
>>>
>>>
>>
>>Hi,
>>
>>please post such question to pgadmin-support or pgadmin-hackers, there is more chance for you to get a rapid answer and others may benefit from such experience :).
>>>I'm trying to connect to a local database with pgadmin and receipt an
>>>error message : src/common/string.cpp(1060) : assert "nLen != (size_t) - 1"
>>
>>Can you give more details please? Which version of pgAdmin are you using? What OS (I suppose Debian GNU/Linux as you asked me first)? What exact release of the OS?
>>Can you configure pgAdmin to log debug information and provide the log file (beware to clean it from personnal informations if you care)?
>>
>>>The database is running in the same machine I'm using the pgadmin III.
>>>
>>So, it may be the pgAdmin 1.3.x release I made during summer with a connection to a local unix socket or are you trying to connect to 127.0.0.1 in tcp mode?
>>
>>
>>>Can you help me ?
>>
>>I bet everybody here would be glad to help :)
>>
>>Waiting for your updates concerning this.
>>
>
> Raphaël,
>
> I'd would like to thank for the help.

NP, but stay on the list please...

> I'm running the Postgres 7.4.2 and the Pgadmin3 1.2.0 Beta 1 (5Nov 2004)
> on a Fedora Core 2 Kernel 2.6.5.

Ok, I've CCed Devrim who is responsible of the fedora port

> The PGadmin3 connection to a remote machine (in other city) occurs OK.
>
> When I try to connect to a database (filling with 127.0.0.1 the field
> Address in the Add server PGadmin form) in the same machine in that
> PGadmin runs too, I receive the message :
>
> src/common/string.cpp(1060) : assert "nLen != (size_t) - 1"

It reminds me an old bug corrected in 1.2.2 and may be in 1.2.0 final
which occured in particular locales when receiving an error message from
the backend.

> Does this have any relation with the problem below ?

I think so (some bet follows)

> I have another problem and I'd appreciate if you can tell me how to solve.
>
> At the shell command line I type : psql -h 200.101.18.22 -d MDI -U MDI
>
> I connect to the database int the remote location (another city) OK.
>
> But if I try to connect to a database that is local instance of a
> database (I dumped the remote database with pg_dump and brought to my PC)
> with the command : psql -h 127.0.0.1 -d MDI_POST -U MDI_POST
>
> I receipt an error message :
>
> psql: não pôde conectar ao servidor: ??????
> O servidor está rodando na máquina "127.0.0.1" e aceitando
> conexões TCP/IP na porta 5432?
> You have new mail in /var/spool/mail/root
> [root(at)Thread data]#
>
> Please, take a look of the blue text above (in English it means : psql :
> can't connct to the server : $#(at)%).
>
> I think the psql is not understanding what -h 127.0.0.1 stands for.
>
> The rule is : only inform -h xxx.xxx.xxx.xxx to remote accesses.
>
> Is this true ?

To me you local PostgreSQL server is not configured to accept TCP/IP
connections.
Can you check the following please:
a) the "postgresql.conf" file should contain the parameter "tcpip_socket
= true"
b) the pg_hba.conf should contain a line like this
host all all 127.0.0.1 255.255.255.255 password
near the beginning of the file (take care that line order is important
in this file).

After that, restart PostgreSQL and check the connection with psql (psql
-h 127.0.0.1 -d MDI_POST -U MDI_POST) and then with pgAdmin... It should
be ok.

HTH,
Raphaël