MSL5026 Hangs

[Rick Dyson]

I have a couple MSL 5026 libraries in different locations, systems, etc. Both have started to exhibit the same behavior where they freeze or hang as far as the robot is concerned. I am still able to access and write to the embedded SDLT drives from the O/S. I just can't control the robot or access the library over the IP connection. Or command the robot over the SCSI path.

One is at v4.23 (current) F/W, the other is a step back or so. I am suspecting network activity of somekind overflowing the web server buffer, etc. Something from a recent worm maybe?

Has anyone else noticed (assuming their library is on an open internet connection) anything similar?

A power cycle clears the problem for a day or so.

rick

[David Ruska]

What is the host/OS?

Are these all SCSI connected, or are some one fibre?

Do you have Compaq Insight Mangement agents running on the servers?

When you lose conneectivity, is the front panel unresponsive as well?

[Rick Dyson]

I use Alpha/OpenVMS v7.3 for both setups with a SAN Switch/FC and MDR for connectivity. I one case, all are fully updated F/W. It has the fewer lockups. In the other case, one has SAN F/W one step back
and MSL F/W one step back.

Both were connected to an IP that is open to the internet (I have no control on that, our institution does not currently have any Firewall setups and we are not allowed to put up any networking equipment... :(.

No Insight, but I do have HP L&TT. I have logged a call on this to CSC, but I was trying to search on my own if anyone else was experiencing the same thing. I believe it to be a side-effect of some DOS, or worm, etc.

rick

[Rick Dyson]

Opps, forgot. Yes, in both cases the front panel is dead to touch. Has to be power cycled with the back switch and then panel touched to turn on.

NOTE: The entire time my VMS Backups run fine as long as they don't need a tape change. My ROBOT commands die with "HW errors".

[David Ruska]

Rick,

I'll ask engineering if the know of any problems having the MSL web interface exposed to the open internet.

Can you light bulb this by disconnecting the web interface, and having it run without failure for an extended time?

[David Ruska]

> Opps, forgot. Yes, in both cases the front panel is dead to touch. Has to be power cycled with the back switch and then panel touched to turn on.

Ok, definately sounds like the library controller is out to lunch.

> NOTE: The entire time my VMS Backups run fine as long as they don't need a tape change.

That makes sense - the drives don't interact with the library exect when loading/unloading the tape.

> My ROBOT commands die with "HW errors".

I'm guessing those are timeouts.

[Rick Dyson]

Yes, both are disconnected from ethernet at this time, and both are OK after 24 hours. :)

rick

[David Ruska]

> No Insight, but I do have HP L&TT.

LTT does not have any polling, so it doesn't have the issue we see with the Insight Management Agents.

> I have logged a call on this to CSC...

If you could provide the case number, I can work them them to have the problem escalated to the engineering team.

[Rick Dyson]

Didn't get to this right away, sorry.

Thanks for the offer of help!

Case # 3205423746.

[Rick Dyson]

I put the ethernet cable back into the library Monday and by Wednesday it was hung again. I had left it out for a week with it being fine the entire time.

Any chance engineering has been able to duplicate this?

rick