Re: Connection to NT Oracle Server v7.3.2.2.1 takes too long

From: Sybrand Bakker <postbus_at_sybrandb.demon.nl>
Date: Wed, 30 Dec 1998 13:09:48 +0100
Message-ID: <368A180C.E457C6C9@sybrandb.demon.nl>

It looks like multiple connection attempts are being made. Two common problems in c:\orant\network\admin\sqlnet.ora are - automatic_ipc = on
This is only useful if you are on the database server. It should be OFF on ordinary clients, if it is ON sqlnet will try to use the ipc protocol with no success.
- names.directory_path = (onames, tnsnames) without an oracle nameserver. sqlnet will try to contact various 'plug and play' nameservers. Either delete the onames entry or swap onames and tnsnames. The best way to isolate this is make sure debugging is on in sqlnet and trace the connection

- set trace_level_client = 16 in sqlnet.ora
- set trace_unique_client to true
- set trace_directory_client to something meaningful (or it will end up in

c:\orant\bin)
This will provide you with a deluge of information, including all sqlnet failed connection attempts.

BTW you probably already have done that, but don't forget to issue a windows nt ping (in a dos box) and a windows nt tracert to verify how long the ordinary windows tcp/ip connect is taking. I have been in a wan where we were connected on the internet directly in Los Angeles. As soon as we needed to consult any Dutch site (including our own) all traffic was directed through the complete US. That was in the same time there was a conflict between the two Dutch backbone providers, both in the same city, and no traffic between them because of the conflict. Very funny.

Good Luck,

Sybrand Bakker, Oracle DBA

"Stéphane Ramet" wrote:

> We have multiple Oracle Servers v7.3.2.2.1 on our Ethernet TCP/IP network,
> both on Windows NT 4.0 Server and WS.
> We access those servers from Windows NT 4.0 WS's.
> In some cases (Client X to Server Y), logging to the Server is immediate,
> whereas in most other cases (Client A to Server B) it takes up to 10
> minutes to get access to the server, using any client tool (sql*plus,
> sqlldr73, etc.).
> It seems that the problem depends on the Cleint's configuration. Any idea
> what parameters we have wrong on our client stations ?
> Thanks in advance for replies.
>
> Stéphane Ramet
> Network Management Tools
> sramet_at_nmg.fr

Received on Wed Dec 30 1998 - 06:09:48 CST