NO_HW on N-class server with 11.11

[Atul Thakur]

We have a N4000-75 class server with 32G of RAM and over 3 Terrabytes of EMC and HP disk over 2 fiber channel cards. One card is the alternate path. The VG's are all defined with the alternate paths.

Problem: This occurred a few days ago. In the middle of the day one LUN came up as NO_HW. The problem was resolved by the EMC person deallocating the LUN using Volumelogix and reallocating it back. The disk was then visible to the server.

Has anyone experienced this problem before ?
Could the OS drop the LUN ?

Any help would be appreciated.

Thanks

[A. Clay Stephenson]

This is not uncommon and the OS didn't lose the LUN. This is most common when using physical disks and one drive becomes defective or uncabled or otherwise unavailable. The error is associated with the device itself (or the connections to the device) rather that the host or the OS. It appears that the configuration data was lost on the EMC array rather than anything wrong with the host itself.

[MANOJ SRIVASTAVA]

Hi Atul


It looks you have a Braocade swtich and lots of luns , as you are using volume logix and lots of space . The reason of the same can vary from the assignmnet of the lun to the switch to the zoning etc lun limitation on the channel which is connected to the EMC .We have a smilar config and have like 8 servers on a SAN swtich , and we have had this problem , W e genrally never had a faulty disk etc , but somehow the assigment in teh swtich got screwed up resulting in faulty Lun in a vg. You need to cehck the following absic things :

1.What is no. of Luns assigned as EMC doenst support more than 256 Luns though it works more than that vlaue.

2.Use basic vlogix , symmetrix command to check the status of the luns atached.

3.Most of the times EMC frames have a inbuilt dialling and disk reallocation incase you dont have these enabled please get this done.

Manoj Srivastava

[Atul Thakur]

Thank you all for the input.

Manoj,
We have less than 256 EMC connections. Also EMC can dial in. When the problem occurred Volume Logix indicated that there was nothing wrong with the LUN and it could see it as allocated, whereas my Unix server came back with a NO_HW for this LUN.


Any new suggestions.

Thanks again.