I/O error

[viks]

hi all..i have got a 12H disk array with two K-360 servers in a cluster.suddenly one volume showed I/O error when bdf is typed.then momentarily the node2 in the cluster gave the error"x controller failed" and it told couldnt access a disk at a specified address.here i have attached the dmesg output.this is an SOS.kindly suggest the possible problem.the dmesg output is of node 2.node1 dmesg output i didnt get.awaiting reply.
thanks in advance.
viks

[Mark van Hassel]

Hi,

It looks as if one of the controllers of the autoraid failed. Because you have Alternate paths to your luns configured the kernel switches to that alternate patchs when the primary patch fails (that's why you see the switch messages). You can check the autoraid with the following commands:
/opt/hparray/bin/arraydsp -i
this will give the serial number which is needed with the next commands:
arraydsp -a [serialnr] # for all info
arraydsp -c [serialnr] # for controller info

Check the status of the controllers. Also, if possible, check the display on the 12H itself.
If there are problems contact HP ...

[Insu Kim]

Did you take a look at the front panel on HP AutoRAID ?
Did it show nothing ?

"Dmesg output" explained that there is a failure in the hardware path for /dev/dsk/c2t1d5 so the ownership of the LUN automatically switches to the alternate path for /dev/dsk/c3t0d5 because alternate links are configured.

For some reason, the problem has been disappeared naturally so the ownership of the LUN for /dev/dsk/c3t0d5 switches to the primary path again.

Dmesg shows this procedure over and over which means that there must be something wrong on AutoRAID.

Check out controller and host bus adaptor !
It's a good idea to issue ARM (Array Management) commands to get a clue as follows:

# arraydsp -i ( To get a serial number )
# arraydsp -a
# logprint -a

Hope this helps.