Well, we MAY have a solution.  The network people made some changes to router
tables last night, and it looks like things might be ok.  I won't be sure until
it all runs cleanly for a few days--they had this fixed once two weeks ago
too--but I think we might have it.

Thanks to all for the suggestions!

Kathleen








Scotty Logan <swl AT stanford DOT edu> on 07/14/2000 11:54:08 AM

Please respond to Scotty Logan <swl AT stanford DOT edu>








 To:      ADSM-L AT VM.MARIST DOT EDU

 cc:      (bcc: Kathleen M Hallahan/ISS/HQ/FHLMC)



 Subject: Re: Losing connection--help!









On 13 July 2000, Kathleen M Hallahan wrote:
> We've been having a problem for the last three nights, and again
> about 2 weeks ago, with the same few servers.  They start to back up
> according to their schedule, and after a certain point (the same
> point every time, generally) the connection gets dropped.  It
> reestablishes successfully, but the process repeats itself ad
> infinitum, unless I manually cancel the active sessions.
...
> -------------------- --------------------------------------------------------
> 07/12/00   18:00:08  ANR0406I Session 13139 started for node BC1NTSPB (WinNT)
>                       (Tcp/Ip 161.107.157.166(1622)).
> 07/12/00   18:09:33  ANR0480W Session 13139 for node BC1NTSPB (WinNT)
>                       terminated - connection with client severed.
> 07/12/00   18:19:06  ANR0480W Session 13155 for node BC1NTSPB (WinNT)
>                       terminated - connection with client severed.
> 07/12/00   18:28:34  ANR0480W Session 13174 for node BC1NTSPB (WinNT)
>                       terminated - connection with client severed.
> 07/12/00   18:38:02  ANR0480W Session 13181 for node BC1NTSPB (WinNT)
>                       terminated - connection with client severed.

We had similar problems recently when a network switch was
accidentally rebooted - several Solaris servers disagreed with the
switch about whether the link was full or half duplex.  The dropped
connections were fairly regularly spaced, as your's appear to be.  The
problems stopped after the switch and servers were forced to
re-negotiate (or the port settings were fixed on the switch).

We still get one or two ANR0480Ws per day, but those are user induced.

  Scotty

--
Scotty Logan - swl AT stanford DOT edu - Unix & ADSM admin - Stanford University
Scotty Logan - swl AT stanford DOT edu - Unix & ADSM admin - Stanford University

"How many times do I have to tell ya... the right tool for the right job!"
  Scotty, Star Trek V: The Final Frontier