Re: session timeout in 10g

well the network guys claim that there is no blocking set up in the iptables.

i enabled sqlnet tracing and found the following entry:

[31-AUG-2005 10:38:00:399] nserror: nsres: id=7, op=72, ns=12586, ns2=0; nt[0]=0, nt[1]=0, nt[2]=0; ora[0]=0, ora[1]=0, ora[2]=0

when i run the trace file through trcassist i get the following:

Error found. Error Stack follows:

              id:7
  Operation code:72
      NS Error 1:12586
      NS Error 2:0
      NT Generic Error:0
      Protocol Error:0
      OS Error:0
      NS & NT Errors Translation

TNS-04315: Trace Assistant Internal error: Can't find resource for bundle oracle.net.trcasst.mesg.TnsError, key 12586

...metalink and google aren't turning much up. any ideas?

On 8/30/05, Mario Cariggi <gelfand.transform_at_gmail.com> wrote:
> Hi Paul, note below
>
> On 8/30/05, Paul Drake <bdbafh_at_gmail.com> wrote:
> > Chris,
> >
> > That sounds like a classis case of attempting to obtain a dedicated server
> > session through a listener where NAT or a firewall is involved. The incoming
> > request to the listener is translated inbounds, but the redirected client
> > never finds its port in the storm. One can tnsping and get a reply, but not
> > create a session.
> >
> > Might you have setup iptables whereby all incoming connection attempts are
> > blocked except for those explicitly allowed? That is a laudable objective,
> > but you'll need to open the high TCP ports (>1024) for dedicated server
> > connections. You can restrict the range of ports used in the kernel
> > settings, if desired.
>
> sorry if I miss something but it seems
> to me that Linux uses shared socket,
> so there is no need to open the high
> TCP ports but 1521: it ain't so?
>
> Mario from Rome.
>

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