HSZ70s & Trucluster

[Mark Poeschl_2]

I'm using a couple older HSZ70 storage arrays to support our dev/test environment. To more closely mimic production we're trying to convert two servers and two HSZ70s to a V5.1B TruCluster (PK4). The revised Release Notes for 5.1B indicate that HSZ70 storage IS supported. Alpha console firmware rev is from the 6.8 CD.

My SCSI HBAs are the differential version of the QLOGIC ISP1040B/V2 and work fine with the HSZ70 in a non-clustered environment. One server's HBA is set to ID 6 at the console (pkx0_host_id) and the other is set to 7.

HSZ70s are running 7.7Z-4 which I'm told is latest and greatest.

All disk units on one HSZ70 are on target ID 0 (D1-D3) and all units on the other are on target IDs 1 and 2 (D100-D104, D200).

I have the whole thing daisy-chained together as a single SCSI bus with hardware terminators on either end.

As soon as one server tries to talk to storage on both HSZ70s at the same time (i.e. copying files from one to the other), they both start complaining about ABORT ERRORS and the copies never complete. Talking to one HSZ70 at a time does not seem to cause any problems.

Is there something about having two HSZ70s on the same SCSI bus that I'm missing? I thought I had everything set up to be conflict free.

[Uwe Zessin]

Mark,
can you please post the output of

> show this_controller full
> show other_controller full

as a .TXT attachment?

[Mark Poeschl_2]

Here is the requested output.

[Mark Poeschl_2]

Ooops! Forgot the "full" the first time...

[Mark Poeschl_2]

Update:
There was an unused SCSI HBA in each server and by making two separate SCSI busses - each one connected to both servers and a SINGLE HSZ70 array - everything works fine.

I'd still be very interested in knowing why multiple HSZ70 arrays on a single SCSI bus caused problems.