hacmp V4.3.1 ip aliasing code

[bmulvey]

Hi,

I have a S80 based cluster with 2 nodes running HACMP 4.3.1
IBM now only support HACMP 4.4.1 so I have to upgrade.

Problem is that we are running a customised bit of code that allows us to use IP aliases for a resource group rather that a service address (thus retaining the boot address at all times). I am not sure that the customisations will work post upgrade.

IBM claim that they will as the IP aliasing became part of 4.4.1 However the documentation is very vague on this.

In order to test this I have 2 spare aix boxes which I installed HA 4.3.1 on and copied all the customisations across. The test cluster starts OK but the aliased Ip address does not start. If I can get this to start, I can then upgrade my test cluster prior to upgrading on the live cluster.

Any ideas or thoughts would be greatly appreciated.

Brian

 

[DSMARWAY]

How are you implementing your ip aliasing ?
what pre or post events have you got your ip aliasing in?

what errors are reported in hacmp.out

what happens if your run your piece of code manually , does it work ?

From what i understand ip aliasing is available under HA 4.5
we have 4.4.1 and i can't find it in there ?

HTH

 

[bmulvey]

Hi DSMARWAY,

Thanks for your response. IBM have come up trumps with the answer.

I got talking to the IBM guy who developed this as a workaround. The workaround works for 4.3.1 and 4.4.1 but got incorporated into the base release of 4.5

As he doesn't develop anymore I won't give his name but he is a rocket scientist.

Below is the text of the Mods to the HACMP files required. Basically he used the MISC_DATA field in the cluster resource definition as a point to trigger his extra code:

Miscellaneous Data [alias%test_cls1/ether1]

I suspect that the actual patch code is IBM proprietary so I better not publish it, just in case, but I guess that a call to CallAIX would get it for a user.

his document then is:

Revised Instructions for installing ipaliasing under HACMP ES 4.4

1. Untar the code into /usr/es/sbin/cluster/local

2. Run the supplied install.sh script to define pre and post events

3. Edit the scripts reconfig_resource_acquire, reconfig_resource_release and reconfig_resource_complete as follows:

First add the environment variables ‘MISC_DATA’ and ‘FS_BEFORE_IPADDR’ to the list of environment variables equated to ‘RESOURCES’
The paragraph should look similar to the following after these additions

RESOURCES="DISK VOLUME_GROUP CONCURRENT_VOLUME_GROUP FILESYSTEM FSCHECK_TOOL RECOVERY_METHOD EXPORT_FILESYSTEM APPLICATIONS MOUNT_FILESYSTEM SERVICE_LABEL HTY_SERVICE_LABEL INACTIVE_TAKEOVER DISK_FENCING SSA_DISK_FENCING TAKEOVER_LABEL NFS_HOST AIX_CONNECTIONS_SERVICES SNA_CONNECTIONS AIX_FAST_CONNECT_SERVICES SHARED_TAPE_RESOURCES MISC_DATA FS_BEFORE_IPADDR"

Next, look in each of these ‘reconfig’ scripts for calls to node_up_local, node_up_local_complete, node_down_local or node_down_local_complete. Change each occurrence of the calls to use ‘clcallev’ to call the script; for example

# Will acquire nfs filesystem in reconfig_resource_complete...
MOUNT_FILESYSTEM=""
export GROUPNAME=$ngroup
node_up_local
# If ANY failure has occurred, this script should exit accordingly
if [ $? -ne 0 ]
then
cl_log 650 "$PROGNAME: Failure occurred while processing Resource Group $ngroup. Manual in
tervention required." $PROGNAME $ngroup
STATUS=1
Fi

Becomes

# Will acquire nfs filesystem in reconfig_resource_complete...
MOUNT_FILESYSTEM=""
export GROUPNAME=$ngroup
clcallev node_up_local
# If ANY failure has occurred, this script should exit accordingly
if [ $? -ne 0 ]
then
cl_log 650 "$PROGNAME: Failure occurred while processing Resource Group $ngroup. Manual in
tervention required." $PROGNAME $ngroup
STATUS=
fi



4. Edit the script ‘swap_adapter’ in /usr/es/sbin/cluster/events as follows:

At the comment ‘# Release SNA resources’ add a line after the 6 lines of comments ‘SNA_CONNECTIONS=’