>Recently, my TSM AIX server has been reporting issues "communicating" with
>our 3494 ATL (plain jane....no VTS).    This is not a new connection. Has
>been working for years.
>
>TSM has started to notice the issues.
>
>I contacted IBM, who says they couldn't find anything wrong.  We checked
>the network cable, end-to-end, with no media issues.
>
>The AIX /var/adm/messages show:
>
>May 19 11:27:01 agena lmcpd[5164]: Timeout polling for read data after
>accept on library 3494atl
>May 19 11:34:24 agena lmcpd[5164]: 3494atl: No CMD response received from
>LM for msg id 11081F7D
>May 19 11:34:25 agena lmcpd[5164]: Taking Library 3494atl offline to host
>and reinitializing
>May 19 11:34:26 agena lmcpd[5164]: Library 3494atl is online to host
...

What I would first pursue, per analogous ADSM QuickFacts entry:

    Query  operation Error - Library is Offline to Host.
        In response to something like 'mtlib -l /dev/lmcp0 -qL' means that
        there is no access because the 3494 is itself offline.  Go to its
        operator station and go to Mode Online.
        For utmost assurance, check that the lmcpd is running, that your
        /etc/ibmatl.conf is correct, and if a LAN connection that within the
        3494 Library Manager you have authorized your host to access it.

If this is a persistent problem, check the 3494 panel for any issues, including
Availability of all components.

If this is an intermittent problem, then look to something like a Denial Of
Service monopolization of the subnet, or its moral equivalent: get your network
people involved.  If no issues there, try shutting down and power cycling the
3494, watching for any problems indicated as it comes up. Perhaps also restart
the lmcpd on your host.  Use ping and periodic mtlib query commands to determine
under what circumstances the outage is occurring.  Observing the 3494 at the
problem times may yield a clue.  If you have lmcpd running on another host in
that subnet, try mtlib commands from there and see if the issue is with your
original host.

If still no results, checking the 3494 internal logs from the panel may yield
some insights.  (The CE usually looks at these - and yours hopefully did - but
anomalies should jump out.)

   Richard Sims