EMS device connectivity errors daily with EMC array

[Gregg Palmer_2]

Hello,
Every day, at about the same time, we receive an EMS error that we've lost connectivity to our EMC CX500 storage array which is connected via 2 GB fibre to our two 9000 servers. Only one box gets the error.
The error states a 'Device connectivity or Hardware error' with a discription of 'the device is not in a state where it can process requests. Manual intervention is required'.
We've gone thru the recommended actions--
We've replaced the adapters and the fibre to the array;
The error messages assume (wrongly) that it is attached via a SCSI bus, and most of the 'probalbe Cause/Recommended action' items really don't seem to apply.
Also, we haven't received any application errors, etc, as a result of this.
Any ideas welcomed.
gregg

[Patrick Wallek]

Run a '/etc/opt/resmon/lbin/set_fixed -L' to see if anything is in the down state. If it is, run a '/etc/opt/resmon/lbin/set_fixed -n device_path_from_previous_command' to reset it.

Now see if you still get the errors.

[Bill Rothanburg]

The EMS disk monitors do not work with the EMC Clariion arrays and should be disabled for your cx500.

HP has an article on how to do this:

http://www2.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000077093864


I use a wildcard to disable all instances down the Fabric Ports where the Clariion is connected.


Bill

[Andrew Merritt_2]

Gregg,
You have an old version of the OnlineDiags; A.38.00 is the HWE0303 release.

The problem with EMC devices being unnecessarily monitored was fixed in the HWE0309 release (A.42.00). CR JAGae75278 covered this.

You should install the latest version of OnlineDiags, which is currently HWE0412, A.47.00, and the latest patch for that version, which is PHSS_32492

You can get the latest version from http://www.software.hp.com/portal/swdepot/displayProductInfo.do?productNumber=B6191AAE

The work-around until you can do this is to use the 'disabled_instances' file (t
he document Bill references describes how to do this). The effect of this worka
round is to prevent any Events being generated for the specified paths; the moni
tor is still actually monitoring the devices.

Andrew

[Saurav_1 · Accepted solution]

Try following,

Check for which LUN is most busy and does it gives any errors in /var/adm/syslog.log

In glance which LUN is being used heavily. Disable bad block relocation on the LV as Array takes care of that automatically.

Please post your error message in EMS and ccheck dmesg For LUN switch or something like that.

Reply.

[Andrew Merritt_2]

Gregg,
Did you upgrade the OnlineDiags, as recommended, and did this fix the problem?

Andrew

[Gregg Palmer_2]

The client will be installing the upgraded OnlineDiags shortly. Thanks to all who replied.

Gregg

[Gregg Palmer_2]

The client is installing the upgrade OnlineDiags to fix this problem...