[Tip] HACMP 7.1.3 and DB/2-Installation

UNIX AIX
1

We just found out that the installation of DB/2 on a HACMP- (PowerHA-) Cluster running v7.1.3 is not possible due to a bug in HACMP. DB/2-version was 10.5 but this seems to apply to 10.1 too (and probably other versions as well).

The error message the DB/2-installer issues is something like:

Do you want to install the DB2 pureScale Feature? [yes/no]
no
DB2 installation is being initialized.

The DB2 installer detected that an RSCT peer domain is online.  Bring the peer
domain offline in order to allow the DB2 installer to update the TSA
installation.  For more information, see the DB2 Information Center at:
"http://publib.boulder.ibm.com/infocenter/db2luw/v10r1/index.jsp".

A major error occurred while installing "DB2 Enterprise Server Edition " on
this computer. The installation cannot continue. If the problem persists
contact your technical service representative.

For more information see the DB2 installation log at
"/tmp/db2_install.log.3277028".

The problem is that the RSCT peer domain is created by the HACMP software during the cluster creation and cannot simply be disabled. You can check for the existence of it with the command

root@system # lsrpdomain

Googling for a workaround I found this in the usenet group comp.databases.ibm-db2, entry by Mr. Greig Wise:

Stop PowerHA and wait till the cluster manager is in INIT state, then run this script:

#!/bin/ksh93
touch /usr/es/sbin/cluster/etc/emgr.$$
clctrl -stop -m <nodename>

but this didn't help us either because the peer domain simply didn't go away.

Solution: before installing DB/2 delete all configured ressource groups of the cluster, then stop it. This will bring down the RSCT daemons, namely ctrmc . After installing DB/2 you can recreate the ressource groups and bring the cluster online again.

PS: Don't bother to report this bug to IBM: that has already been done.

I hope this helps.

bakunin

2 Likes
2

Hi,

i found some hints in my RMC/RSCT documentation but i dont know if them would help in your case:

RSCT Peer Domain Commands:
mkrpdomain - Create a peer domain
rmrpdomain - Remove a peer domain
lsrpdomain - List characteristics of a peer domain
startrpdomain - Bring a peer domain online
stoprpdomain - Bring a peer domain offline

Stopping ctrmc will not stop the ressource managers.
For proper restart use:
rmcctrl -z
rmcctrl -A

Regards

1 Like
3

I had a lengthy reply - but it never made it out - sigh.

Just note that RSCT activity is normal, even necessary, to have DLPAR aka Dynamic Resource Management working.

root@x062:[/]oslevel -s
7100-03-05-1524
root@x062:[/]lssrc -a | grep rsct
 ctrmc            rsct             6160584      active
 IBM.HostRM       rsct_rm          6947044      active
 IBM.DRM          rsct_rm          7078142      active
 IBM.MgmtDomainRM rsct_rm          6684680      active
 IBM.ServiceRM    rsct_rm          5963970      active
 ctcas            rsct                          inoperative
 IBM.ERRM         rsct_rm                       inoperative
 IBM.AuditRM      rsct_rm                       inoperative
 IBM.FSRM         rsct_rm                       inoperative
 IBM.SensorRM     rsct_rm                       inoperative
 IBM.ConfigRM     rsct_rm                       inoperative
 IBM.WLMRM        rsct_rm                       inoperative
 IBM.StorageRM    rsct_rm                       inoperative
 IBM.LPRM         rsct_rm                       inoperative
 IBM.MicroSensorRM rsct_rm                       inoperative

As much has changed in RSCT between the different versions of AIX 5.3 and AIX 7.1 (soon to include AIX 7.2) - your mileage may vary. In any case, ctrmc is supposed to be active all the time. And for years, the cluster manager demon that PowerHA installs is supposed to be active - all the time (in state ST_idle perhaps!).

For a lot of bla bla about what ctrmc is doing try:

root@x062:[/]lssrc -s ctrmc -l

My experience with DB2 and installs on AIX is that DB2 (read software group) does add specific stuff for AIX (they use neither installp nor rpm for their install method) - and when AIX changes enough that there modification/specification no longer fits - install may fail, or - more frequently - a warning message that is no longer applicable.

If the install is "uninstalling" itself, rather than just complaining check around in the sub-directory (was named db2_setup if I recall correctly) and see if you can find the test, change the logic (or comment it out) - and see if it works.

Please note - the last time I was doing modifications like this it was still db2 version 9.something.