trace AUTOSPARE disk on HSG80 controller

[Matt Palmer_2]

Hi,

I have a mirrorset in which a disk failed overnight, and one of my spareset disks jumped in. The only problem is, I cant tell which one!!

All I know is that the original disk has been moved to the failed set. Is there any way I can tell where a spareset disk has moved to or where a failedset disk has come from, as I am anxious to know which mirrorset had the problem.


I know I could just replace the failedset disk, and make that the new spare, but I would prefer to be able to track the locations of my storage.

any help would be great

thanks

Matt

[Michael Schulte zur Sur]

Hi,

do you have a documentation of the location of the spare disks before the failure of the disk or were the mirror sets in a symmetric fashion in the enclosures?

What host and os version are you running?

Michael

[Enemy]

It is good practice to label your disks/units in storage :) Did you know witch disks where in spareset before Failure? Then use:
HSG80>show spareset
HSG80>show failedset
You can use CLI's LOCATE comand to physicaly locate you disk in storage (disks will flash failure LED)

[Matt Palmer_2]

unfortunately, I inherited the storage, so I have only mapped my additions. So I am able to track any of my disks as I did label these ones.

I have spoken to HP and they have said that there is really not an easy way to reverse trace the swapping-in of the spareset, unless you catch the spareset replacement whilst the mirror is normalizing.

thanks for the advice though.

[Michael Schulte zur Sur]

Hi,

as enemy suggested you can locate each mirror set by locate mirrorname and see, if there is any irregularity in the positioning. Othewise if you have a monitor attached to the console port of the hsg80, you may find the event on screen. But other than that I don't know. So for the future make a printout of:
show disk full
show spare
show mirror full
show raid full

Michael

show unit full

[Michael Schulte zur Sur · Accepted solution]

Hi,

what I meant with my last posting was, maybe your predecessor used a certain pattern to place the mirror sets and if the mirror set with the replaced drive look different, then you have the former spare drive.

Michael

[Uwe Zessin]

You can use the LOCATE {object} command which helps to visualize where a disk is located or which disks belong to which mirrorset and so on. Just remember that the LOCATE is 'cummulative(sp?)'. If you say 'LOCATE M1' it displays all disks that belong to 'M1'. If you then say 'LOCATE M2' it will display the disks of M1 _and_ M2. You can clear with 'LOCATE CANCEL'.

I also suggest to try a 'LOCATE ALL' after any changes have been made - I once had a disk drive enclosure where one bay was missing all LEDs!

--

I have just checked, but unfortunately disk replacements seem not to be logged in the last_failure array of the controller:

HSG-A> run fmu

Fault Management Utility

FMU> show last_fail all

[Kenneth Ray_4]

Uwe,
How do you get out of the FMU utility? After I type 'run fmu' I can't get back out of FMU. I try exit and it exit's me out of the whole controller and when I connect again it still has me in the fmu.

[Uwe Zessin]

Kenneth,
how (which way) did you connect to the controller?

[Kenneth Ray_4]

Uwe,
I connected to the controller via hsxterm5 on a tru64 server.

[Uwe Zessin]

I see... It looks like the program intercepts the input from you and interprets it first. Well, I don't have experience with 'hsxterm5', but I would try this:

- connect in interactive mode and send a CONTROL-C or CONTROL-Y

- send the 'exit' command with some space characters surrounded, like " exit " in the hope that hsxterm5 can be fooled that easily.

- send it as a one-time command like:
# hsxterm -fccl_device "exit"
or in variations like " exit" or " EXIT"

- put the string 'exit' in a file and send it as a script:
# hsxterm -fccl_device -sexit.txt -lexit.log

If that does not work I am afraid you have to restart the controller to get rid of the program.

[Mohamed K Ahmed]

Connect through the serial port in the back of the controller, it is much easier.
Then you can run whatever you want of tests and you can see the drive's leds while working

Mohamed

[Uwe Zessin]

The serial port is at the front of the controller module (the back is the side with the connectors which go into the BACKplane), but it depends on the enclosure type if the controller is sitting in the front or the back of the enclosure.

If you have a BA-370 enclosure, then the controllers go in at the front and you can easily see the disk drive LEDs.

If it is a modular array with an M2200 enclosure, then the controllers stick in the back and you have to walk round the cabinet or route the cable to the front so that you can see the disk drive LEDs.

[Michael Schulte zur Sur]

Hi Kenneth,

once I fell into that trap myself. When you type exit, you are on the shell again. :-(
I solved the problem with the cli of the storage works command console. There I could type exit leave the fmu. Of course using the serial link as suggested is the same.

hth,

Michael

[Mark S_3]

I just got myself stuck in hsxterm5 in the FMU. To exit I proceeded exit with v. Worked for me.