[derek]

Hi Guys,

Running out of ideas on solving a DLPAR problem we are having.

Some Background..

Running a large environment with around ~200 servers, most running HA.

It is a mixed environment made up of mostly p590/595 and p690, so in turn we also have a mix of both POWER5 and 4 HMC's

All HMC's and AIX servers are setup with DNS that resolves short and long hosts to IP's and also reverse.

We have DLPAR working only on a handfull of nodes, and they can change aswell, one day one will be dlparable, one day it won't etc..

All /etc/ct_node_id files are unique, I have tried rebuilding (reseeding - ctsthl -s) the trusted host lists so that /usr/sbin/rsct/bin/ctsvhbac shows all RSCT interfaces show up as Trusted Hosts. All this done inside stoping/starting the daemons and different combinations aswell.

I have followed every step in the DLPAR Checklist atleast twice (read: Dynamic LPAR tips and checklists for RMC authentication and authorization)

I have probably missed a few things I have tried here but any help would be reatly appreciated.

HMC Levels.

POWER5 = 6.1.2

POWER4 = 3.3.5

Oslevels = 5.3-5 and 5.2-6

Thanks in advance

Derek

[derek]

Well it appears I may have solved the POWER5 HMC's DLPAR

I noticed that the HMC's Open network adapter eth1 did not have the "Partition Communication" box checked which enables the RMC / RSCT communication to LPAR's on this interface. I believe this could have happened after a recent upgrade to 6.1.2 from 5.2. Still unsure if it has worked as it needs to be approved for change etc..

Still the POWER4 HMCs do not have this option, I guess this is because they use a serial link method to connect to the managed systems and it assumes that the ethernet interface is used for partition communication.

FYI - this problem has only occured in the last 5-6months - DLPAR on both HMC models were working before this