Accessing disks in MA8000 with hsxterm5

[Derk Korevaar]

Hi,

We have a storage bay with several systems connected. From Tru64 V4.0G we can get access to the MA8000 controller with:

/usr/lbin/hsxterm -f /dev/rzh137c

On the systems running Tru64 V5.1B, I have /usr/lbin/hsxterm5 but when I try to use it I get: ScsiExecCli Failed.

What am I missing?

Thanks,

Derk Korevaar

[Michael Schulte zur Sur · Accepted solution]

Hi,

the disks get a new name under 5.X.
please run scu show edt and use the syntax
hsxterm5 -b X -t X -l X
where you have to substitute X with the parameter shown from scu.

greetings,

Michael

[Hein van den Heuvel]

Hey Derk! Fancy meeting you here !

Yeah, forget the rz... devives.
Think: /dev/rdisk and dev/disk but you found that right?

Did you find the /dev/cport/scp devices?
That is really the way to go.
Do not talk to the controller through a disk link, but define a controller/command lun.
[There actually have been report of AdvFS corruption when talking to the controller through a disk]

Check with hwmgr -show scsi
See that scpX ? See the same Bus and Target, but Lun=0? For example:

:
94: 12 x disk none 0 1 dsk10 [2/4/3]
95: 13 x disk none 0 1 dsk11 [2/4/4]
96: 14 x disk none 0 1 scp1 [2/2/0]
:

[Hmm, I am working with an HSG on this system, the HSV behaves the same right?]


Groetjes,
Hein.

[Derk Korevaar]

Hi Hein,

We have a slight problem: our V5 system is connected to a bay with V4 systems, so the bay is at SCSI 2 with no CCL:

root(jade):/ # scu sho edt

CAM Equipment Device Table (EDT) Information:

Bus/Target/Lun Device Type ANSI Vendor ID Product ID Revision N/W
-------------- ----------- ------ --------- ---------------- -------- ---
0 5 0 Sequential SCSI-2 COMPAQ SuperDLT1 3434 W
1 0 0 CD-ROM !ANSI HL-DT-ST CD-RW GCE-8320B 2.04 N
5 0 10 Direct SCSI-2 DEC HSG80 V86S W
6 0 10 Direct SCSI-2 DEC HSG80 V86S W
7 0 0 Direct SCSI-2 COMPAQ LOGICAL VOLUME 2.88 N
7 32 0 RAID SCSI-2 COMPAQ Smart Array 5300 2.88 N

root(jade):/ # hwmgr -sho scsi

SCSI DEVICE DEVICE DRIVER NUM DEVICE FIRST
HWID: DEVICEID HOSTNAME TYPE SUBTYPE OWNER PATH FILE VALID PATH
-------------------------------------------------------------------------
72: 1 jade cdrom none 0 1 cdrom0 [1/0/0]
73: 2 jade disk none 2 1 dsk0 [7/0/0]
76: 0 jade tape none 0 1 tape0 [0/5/0]
77: 3 jade raid none 0 1 scp0 [7/32/0]
91: 4 jade disk none 2 2 dsk1 [5/0/10]
92: 5 jade disk none 0 2 (null)

Both /usr/lbin/hsxterm5 -b7 -t32 -l0 and /usr/lbin/hsxterm5 -b7 -t0 -l0 will conect me to the internal bay with mirrored bootdisk (result ScsiExecCli Failed). For the MA8000 my only choices are /usr/lbin/hsxterm5 -b5 -t0 -l10 or /usr/lbin/hsxterm5 -b6 -t0 -l10.

I hope we don't encounter an AdvFS problem...

Regards,

Derk Korevaar

[Uwe Zessin]

I thought that 'hsxterm' is not supported on HSG controllers. The replacement is the 'HP StorageWorks Command Scripter' which, unfortunately, costs money.
http://h18000.www1.hp.com/products/sanworks/commandscripter/index.html

Hein,
the EVA always run in what is known as 'SCSI-3 mode' on the HSG. It always has a LUN at address 0, but a host cannot talk that way to the EVA. You need to get the platform kit for your OS and install SSSU (storage system scripting utility). This one will talk via TCP/IP to the Storage Management Appliance (SMA) which will send your request to the EVA.

[Michael Schulte zur Sur]

Uwe,

hsxterm5 works on hsg80.

greetings,

Michael

[Uwe Zessin]

Michael,
I did not say that it does not work. It is not unusual that things that are not supported happen to work. And the opposite can happen, too :-(

Hein has already written that this can cause ADVFs panics. It is also possible that a combination of serial and in-band access can cause a controller to freeze. Nobody knows what else can happen, because it has not been thoroughly tested.

[Michael Schulte zur Sur]

Derk,

what do you want to do with the hsxterm5?
Do you have the cd for the hsg80?
You could use the swcc for the configuration.

greetings,

Michael

[Derk Korevaar]

Hi,

Some more details:
- we have swcc installed, so usually we manage using swcc on a PC. But it is not always convenient, so we often access the HSG80 direct using hsxterm on Tru64 V4

- We rarely access the bay from our system running Tru64 V5, so I was not aware that the syntax had changed between V4 and V5

- I hope we don't meet AdvFS panics, as a CCL is not available. We are running in a mixed V4-V5 environment, using SCSI-2 and no CCL (V4 does not like SCSI-3 and a CCL at lun 0 would mean re-configuring our disks in LSM as unit 0 would not be available any more)

- We now have a new customer, running Tru64 V5 but using an MSA1000 rather than a HSG. I get the feeling that the MSA1000 can only be managed using the console port. hsxterm5 does not work.

Regards,

Derk Korevaar

[Uwe Zessin]

Derk,
you are 'skating on thin ice'.

The HSG requires an interlock if it is feed commands from different agents - otherwise it can lock up. Old versions of SWCC do not know how to take that lock.

--

You observation about the MSA1000 is correct. The command set is (somewhat) modelled after the HSG CLI, but it cannot be fed commands with HSXTERM or the Command Scripter I have already mentioned.

The OpenVMS operating system comes with a program called 'MSA_UTIL.EXE' that allows the user to talk to the controller through the MSA's "CCL", but some initial setup need to be done through the serial port, too.

I haven't seen such a program for Tru64 Unix, yet. And I haven't seen a version of ACU for the AlphaServers, either.