Eduardo:
I have just completed setting up a two-node hacmp cluster in cascading
mode with hardware address takeover and ipaddr takeover running Netview 5.1
and the Tivoli framework 3.6 on the primary host and Informix on the
secondary host. I am using AIX 4.2.1 and HACMP 4.2.2 with all shared
filesystems on RAID1 external scsis model 110.
In general the install and configuration is not too difficult if you
know a couple of tricks such as:
(1) Netview checks the hostname to determine if it is on the proper
machine.
(2) Tivoli checks the uname -S to determine if it is on the proper
machine.
(3) Informix checks the ipadapter lablel to determine if it is on the
proper machine while the Informix snmp subagent checks that the ipadapter
label resolves to the hostname.
The hostname, ipadapter labels and unames must be set correctly in your
startup and shutdown scripts for your resource groups that netview and
tivoli belong to in order for failover to work correctly.
(4) After you install the framework and netview on one node using, say,
two different shared filesystems on the same external disk such as
/dev/lvnetview=/usr/OV and /dev/tivoli=/tivoli, then you varyoff the volume
group they belong to, unmount the filesystems, varyon the volume group on
the other node, mount the filesystems and then delete all files in them
before doing the install all over again. If you don't do this, the install
programs will think there is not enough disk space.
There are also a couple of updates to the AIX bos you must have installed
for HACMP to work correctly, such as one for X11 or the xhacmp gui cores.
In addition all disk adapters, disk controllers and scsi controllers must
have their microcode upgraded for HACMP to work correctly, get ECA190 and
ECA010.
Other than these things, it's a breeze.
good luck,
klg
|