Modular Data Router not recognized in windows device mgr

[CA1422843]

I have a Compaq Storage works Modular Data Router connected to the SAN. The MDR connects to 4 MSL5000 Tape libraries via SCSI. I have 7 Win 2003/2000 servers running CA Brightstor ARCserve and they connect to the MDR and use the libraries to back up data. The problem i have is communication errors between the libraries and the servers. In the servers device manager the MDR is shown with a yellow ? mark and no drivers are loaded. Do i need to have drivers loaded for this to work properly?

Any help would be greatly appreciated!!
Richard

[Urban Petry]

You don't need any driver for the MDR.
There is however a sort-of "dummy" INF file provided by HP (at least in the past ...) so the device does not show up as an unknown device - the driver itself does nothing (it's Microsoft's "null.sys").

Do you use SSP (selective storage presentation)? I would strongly recommend to manually configure it (as well as static mappings on the Windows server's HBAs) to be sure that intermittent device malfunctions don't mess up the device order as they're seen by the windows servers.

Urban

[CA1422843]

Thanks Urban,

I not familiar with SSP, so i am not sure if we use it or not, i'll have to find out. I havent checked the device order as seen in the windows servers, i'll take a look at that.
When i look in the tape.log for ARCserve it says things like "library thinks there is a tape in slot 1 but no tape found" But there actually is a tape in slot 1.
I also have problems opening the library doors becuase "a SCSI host is controlling the device" when nothing is running on any of the servers.
Right now i am still verifying that all the hardware is on CA's Certified Device List, so far I have verified everything except the SAN switches (still cheking on those).

[Urban Petry]

Richard,

have you tried the "Library & Tape Tools" (L&TT) software from HP to access the library and/or drives in case the backup application can't access them anymore? This can give you a clue as to whether the problem lies within the backup software or "down the SAN".

Try to access the library and tape drives with L&TT from the same server ARCserve is reporting the error on.

Urban

[CA1422843]

The L&LT software connects to the libraries with no problems, it shows the tapes i have in it and everything.

[Urban Petry]

Then the problem is on or above the OS level.

Do you have those problems all the time or is there a situation when everything works for a while (e.g. after reboot of all servers, MDR and libraries)? In the latter case do you have any event log entries when the problem starts (device removals, SCSI errors)?

When you experience the problem can you reconfigure the backup devices within ARCserve and test again whether this helps make it work again (I haven't used ARCserve for years so I can't tell you more precisely what to do; just repeat the same that you did when configuring the devices for the first time :-).
If after this reconfiguration everything works again (at least for a while) I bet the problem is caused by device reordering due to SAN problems or other temporary backup hardware problems.

Urban

[CA1422843]

I have reconfigured the device groups in ARCserve but i dident notice anything positive, however i did notice that when one of the libraries was down for a bad PS for a few days when it came up again it worked perfect for about 2 days. So i tried this on the other libraries, shut them off for about an hour and it would work good for a few hours.
ALso on the SAN - it is set up with redundancy, so each server has 2 HBA's plugged into a pair of fabric switches, the problem is the MDR only has 1 fiber connection so it only plugs into one of the fabric SANs.

[Urban Petry]

Hmm, very strange ...
With that information the problem might not be related to OS at all but to the MDR/library.
After switching off one library for an hour is more than one server accessing it successfully before the problem starts again?
If the problem reoccurs can you successfully see the library's contents via L&TT from _all_ servers (that would rule out SCSI reservation conflicts)?

Urban

[CA1422843]

Yes more than one server can access the libraries okay after the 1 hour shutdown period.
Also it is not a matter of IF the problem reoccurs, it's when :)
I can access the libraries fine with the L&TT tools.
Also we have an extra MDR and i have switched to that one to eliminate it as a possibility.
Once the problem begins it does not stop, some backups will run some wont. From what i can tell of the tape.log (error log for ARCserve) it seems that some of the commands are not getting through or they are garbled or mixed with other commands.

[Stephen Kebbell]

Hi,

what else do you have in the SAN? Do you have disk and tape devices zoned separately?

Regards,
Stephen