Re: Rookie Error?

Thanks, Jared.

comments in-line...

On Sun, 28 Nov 2004 20:21:08 -0800, Jared Still <jkstill_at_gmail.com> wrote:
> Sure, there here are in order of not working, and working.
>
> [root_at_sherlock tmp]# netstat -an | grep 1522
> tcp 0 0 127.0.0.1:1522 0.0.0.0:* LISTEN
> [root_at_sherlock tmp]# netstat -an | grep 1522
> tcp 0 0 0.0.0.0:1522 0.0.0.0:* LISTEN
> tcp 0 0 192.168.1.101:1522 192.168.1.105:39986 TIME_WAIT
>

IIRC this machine is White Box (similar to RH Enterprise 3) and the closest reference I could find to the solution was in this metalink document (and more generally in RAC docs): http://metalink.oracle.com/metalink/plsql/ml2_documents.showDocument?p_database_id=NOT&p_id=287453.1 (see section #4 for /etc/hosts configuration and other goodies specific to RHEL3 and SUSE Enterprise)

There must be something different with RHEL3 than Fedora 1 for example. I tried to simulate your problem on Fedora 1:

/etc/hosts (supposedly wrong but allows remote connection) 127.0.0.1 localhost fe1-60.tec.com fe1-60

[root_at_fe1-60 root]# netstat -an | grep 1521

tcp        0      0 0.0.0.0:1521            0.0.0.0:*               LISTEN
tcp        0      0 127.0.0.1:1521          127.0.0.1:44592         ESTABLISHED
tcp        0      0 137.27.40.60:44524      137.27.40.60:1521       TIME_WAIT
tcp        0      0 127.0.0.1:44592         127.0.0.1:1521          ESTABLISHED

/etc/hosts (corrected, also allows remote connection)
127.0.0.1            localhost
137.17.40.60      fe1-60.tec.com fe1-60

[root_at_fe1-60 root]# netstat -an | grep 1521
tcp        0      0 0.0.0.0:1521            0.0.0.0:*               LISTEN
tcp        0      0 137.27.40.60:44524      137.27.40.60:1521       ESTABLISHED
tcp        0      0 137.27.40.60:1521       137.27.40.60:44524      ESTABLISHED

The way I interpret these stats (correct me if I'm wrong) is that the 0.0.0.0 means the logical host, and the .0 address suffix typically means a network segment vs NIC (NIC has 1-2XX).

I now understand. The problem occured because the local-loop (similar to NIC) address 127.0.0.1 was LISTENing to port 1522. In both my cases with Fedora 1 the 0.0.0.0 (logical host) address was LISTENing, and I could not (figure out how to) reproduce the problem. Its seems obvious now that I know what to look for. ;-)

I'm sure people with RAC experience consider this trivial, as well as systems with many NIC cards and network segments. I've summarized it so I can 'hammer it in my brain' and get feedback if I'm off track.

Thanks to MZ for the solution and explanation, and Jared for patience.

HTH Regards,

Mike Thomas

--
http://www.freelists.org/webpage/oracle-l