rx8620 pci power module failure

[Alex_R_1]

If this fails should the system run in a degraded state? Is this a single point of failure piece of hardware?

If it should/can run in a degraded state is there a configuration that needs to be put in place?

[TTr · Accepted solution]

Look at the quickspecs and read the "AC/DC Power" section, "PCI Power Supplies" paragraph. It appears to be a SPoF for the I/O bay. http://h18000.www1.hp.com/products/quickspecs/11849_div/11849_div.HTML#Technical%20Specifications

[Alex_R_1]

Thanks, that is exactly what I needed. Looks like I need to look in to having Service Guard not only fail over between nodes but also between nPar's.

[Torsten.]

Serviceguard between nPars is a bad idea ... consider to use independent systems.

[Alex_R_1]

Why do you say that? Do you have any reading material to support the statement? Have you had real life experiance with it?

[Torsten.]

A cluster in a single cabinet is not high available.

Think about planned or not planned events like hardware failures, maintenance, air condition and power loss, etc. etc. ...

[Alex_R_1]

I agree with you statement. The cluster will not be removed. But the system should have failed to a degraded state.

Service Guard did fail over properly to the other node.