dead1e c0ffee

The front panel of the HMC has three buttons ( ) below it. When the machine is booted up, it shows a few lines of system information.  It has two modes; normal and maintenance.  In normal mode, you have one of 4 possible menus that you can display.  They are labelled 1,2,4 and 20.  You press the enter button  to toggle between displaying information and changing menus.  Menu 1 is an overall status, menu 2 allows you to change the service processor you want to work with (A or B, or just A) and also change the operating mode (N or M). In maintenance mode, you have menus labelled 1,2,3,4,7,8,20,25,26,30,42 and 43.  We are only interested in menu 30, which will show us the two ip addresses configured for the HMC to use.

First, put the HMC in manual mode.

01 N V=F HMC=1

T

Push Enter

01

Push Up Button

02

Push Enter

02 A<N

T

Push Enter

02 A N<

T

Push Up Button

02 A M<

T

Push Enter

02

(You are now in Maintenance Mode, if you were to go back to the home screen by going to 01 and pressing enter, you would see the M where the N used to be)

Next, go to section 30:

Push Up Button until you get to 30

30

Push Enter

30**

Push Up Button

3000

Push Enter

Push Down Button

3000

Push Down Button

3001

Push Enter

When you are done, you should probably change back to normal mode, go to menu 01, and press enter to leave the display showing what we see at the top of this post.

The hostname gets hardcoded into a couple of nasty places that the ‘hostname’ command doesn’t see or change.  The first is ‘uname -n’ which you set with ‘uname -S newname’.  Oracle tends to use old perl scripts that use this to find the hostname.  To be safe, read the rest of this post also.

The hostname command itself seems to simply read from inet0, but sometimes this doesn’t work:

lsattr -El inet0 | grep hostname

chdev -l inet0 -a hostname=newname

Apparently if you change the hostname of a system, it doesn’t get changed in the IBM.Host stanza of rsct, which CRM relies on.  You must use a refresh command to change this.  Also, if your hostname seems to magically change back to on old name on reboot, check /etc/rc.net to see it is hardcoded there.

> lsrsrc IBM.Host
Resource Persistent Attributes for IBM.Host
resource 1:
Name = “newnodename”
NumProcessors = 16
RealMemSize = 5033164800
OSName = “AIX”
KernelVersion = “5.3″
DistributionName = “IBM”
DistributionVersion = “5300-06-07-0818″
Architecture = “ppc”
NumOnlineProcessors = 8
EntProcCapacity = 150
NumOnVProcessors = 4
NumActPProcessors = 4
ActivePeerDomain = “”
NodeNameList = {“oldnodename”}
> /usr/sbin/rsct/install/bin/recfgct
(doesn’t return anything but takes a few seconds to regenerate from the HMC)

> lsrsrc IBM.Host
Resource Persistent Attributes for IBM.Host
resource 1:
Name = “newnodename”
NumProcessors = 16
RealMemSize = 5033164800
OSName = “AIX”
KernelVersion = “5.3″
DistributionName = “IBM”
DistributionVersion = “5300-06-07-0818″
Architecture = “ppc”
NumOnlineProcessors = 8
EntProcCapacity = 150
NumOnVProcessors = 4
NumActPProcessors = 4
ActivePeerDomain = “”
NodeNameList = {“correctnodename”}

So to recap:

hostname newname
uname -S newname
chdev -l inet0 -a hostname=newname
/usr/sbin/rsct/install/bin/recfgct
check /etc/rc.net to see if the hostname has been hardcoded into it by HACMP or something

Sometimes when you try to DLPAR something from the HMC, it fails to communicate, it is because some daemons aren’t running on the client.

I have run this on a vio server. I don’t know if it will affect a client that is already using rsct for something else (HACMP):

to reset the deamons
usr/sbin/rsct/install/bin/recfgct
/usr/sbin/rsct/bin/rmcctrl -p
/usr/sbin/rsct/bin/rmcctrl -z
/usr/sbin/rsct/bin/rmcctrl -A