replacing FC adapter

[ogniemi]

please tell me the right steps for HBA replacing in HACMPed environment. ESS + sdd.

Lets say, fcs0 is failed. When I stop hacmp services on affected node are these steps enough and ok?

1. rmdev -dl fcs0 -R
2. shutdown node
3. replace adapter in box (it is p55A)
4. power on
5. -- I expect new fcs0 devices are configured, right?

I there anything else to be done?

Should vpath devices be offline on all cluster nodes or I can keep hacmp services / resource group running on backup node?

Will reservation kept on LUNs on backup node have impact during configuring new adapter on the node were it will be replaced? Will cfgmgr configure vpath/paths properly on affected node when LUNs are reserved on backup node?

 

[khalidaaa]

IMHO I don't think this will affect the backup node in any way. But you would have to visit your ESS switches and recofigure your zoning to include that new fcs in place of the old one.

Regards,
Khalid

 

[dial8d]

Presumably you have 2 fibre adapters in your active cluster node. If so, you could hot swap the failed adapter without taking down the service, or failing the service over.

 

[ogniemi]

as far as I know P5 9133-55A has't hot swappable PCI slots so I think server has to be powered off to replace the adapter.

san switches are already prepared for new HBA.

 

[khalidaaa]

According to that hopefully you shouldn't face any problems!

Regards,
Khalid

 

[p5wizard]

I'd go in diag and check the slots, I believe in p5-55A they are hotswap...


HTH,

p5wizard

 

[dial8d]

they are indeed hot swappable.

lsslot -c pci

This shows you the hit swap pci slots on your system.