Changing failing cards in npar

[Waqar Razi]

If some hardware internal to the server lets say dimm fails on an npar and we want to replace it, do we need to shutdown the whole complex for the replacement or just that particular npar?
Can some one please give me steps to change any failed hardware component in an npar?

[Robert-Jan Goossens]

Hi Wagar,

Have a look at the olrad command.

---
The olrad command provides the ability to perform on-line addition and replacement of I/O cards.
---

I don't think you can on-line replace a dimm on a HPUX system.

Regards,
Robert-Jan

[Matti_Kurkela]

Check the MP/iLO of the nPar system. If it can independently power down the component you're trying to replace (or the component it's attached to, e.g. a cell board), you can replace that component without powering down the entire complex.

MK

[Michal Kapalka (mikap)]

Hi,

if you are using Npar, you need to shutdown the Npartiotion and them replace the memmory.

If the memmory make a single bit error, it will be ok, but if it's a multiple bit error,
the system will crash.

Michal

[Torsten.]

This all depends on what you have and the way it is configured.

If some (or better many) requirements are met you can even replace a DIMM or CPU "online" (by removing a cell from a running partition).

But you should always ask HP for support for your partitionable hardware.

[Torsten.]

Just to add

"If the memmory make a single bit error, it will be ok, but if it's a multiple bit error,
the system will crash."

This should not happen on a sx2000 based system - even a multibit error should be corrected.

[Sherif A. Louis · Accepted solution]

Wagar, depending on which generation of chipsets you have, yosemite, sx1000 or sx2000 and which operating system HP-UX 11i v3 or prior? Also depends on how the npar is configured. If it's a core cell for example then you will definitely have to shutdown the whole partition. If it's not then u can simply do a
# parmodify -p partition # -d cell#
which removes the cell from the npar
from the MP shutdown the cell using "pe" command
get out your cell change your dimm, get it back in power it up. added to the partition.

if you don't want to mess with your complex profile then you will have to shutdown your whole partition and this is if you don't have a system that support dynamic npars

If you want to change an I/O you can do that online using the olrad command under HP-UX. It doesn't work on linux if it's an integrity.

In the end it's better of course to have an HP support engineer to do this for you.