SEO

#1 (**permalink**) 01-05-2008, 07:39 AM

Hi
I configured 2 AIX nodes and installed OS 5.3-ML 2 / HACMP 5.2 on
each. I configured one ethernet interface on each node. I want to get
rid of extra tcpip card usage by configuring 2 diskhb interface on each
node. Is it OK ?

I configured service ip . Service IP is bound to ethernet network.

I started hacmp service on one node..Vgs varied on and files mounted
properly. I started hacmp service on second node. Hacmp services
started on Second node and it got shutdown automatically after 1-2
minutes. I need help to resolve this issue.

Appreciate

MNI

#2 (**permalink**) 01-05-2008, 07:39 AM

MNI wrote:
> Hi
> I configured 2 AIX nodes and installed OS 5.3-ML 2 / HACMP 5.2 on
> each. I configured one ethernet interface on each node. I want to get
> rid of extra tcpip card usage by configuring 2 diskhb interface on each
> node. Is it OK ?
>
> I configured service ip . Service IP is bound to ethernet network.
>
> I started hacmp service on one node..Vgs varied on and files mounted
> properly. I started hacmp service on second node. Hacmp services
> started on Second node and it got shutdown automatically after 1-2
> minutes. I need help to resolve this issue.

I don't think this will work. As far as i know you need
at least two IP networks (4 NICs in a two-node cluster)
and one non-IP Network (serial or disk based) for HACMP
to work. These prequisites are given by the way HACMP
determines if there's only a local network failure (just
switch cards) or if the remote node is actually down
(switch services to other node).

BTW why would you want to run a cluster with only one
NIC per node? Even SunCluster, which doesn't need the
non-IP network, won't run with this kind of config.
Actually i can't think of a situation where this kind
of setup makes sense! Am i missing something here?

Regards,

Frank

#3 (**permalink**) 01-05-2008, 07:39 AM

MNI wrote:
> Hi
> I configured 2 AIX nodes and installed OS 5.3-ML 2 / HACMP 5.2 on
> each. I configured one ethernet interface on each node. I want to get
> rid of extra tcpip card usage by configuring 2 diskhb interface on each
> node. Is it OK ?

I think you MUST configure at least one network for heardbeat so HACMP
can determine which node is right now acting as the 'primary node' so
it needs to communicate with the other node. Heardbeat over harddisk is
NOT a kind of morse.

Please read first the books and requirements about HACMP. Otherwise you
do nobody good.

hth
Hajo

#4 (**permalink**) 01-05-2008, 07:39 AM

HI
Thanks all for your suggestion..I was so for using 3 tcpip [2 for
network heartbeats] cards on each node. We are running short of NICs
for a new 2 node cluster setup, thats why I want to give a try.

Thanks for all your help.

MNI

#5 (**permalink**) 01-05-2008, 07:40 AM

MNI wrote:
> Hi
> I configured 2 AIX nodes and installed OS 5.3-ML 2 / HACMP 5.2 on
> each. I configured one ethernet interface on each node. I want to get
> rid of extra tcpip card usage by configuring 2 diskhb interface on each
> node. Is it OK ?

Yes it can be made to work but if you must consider and plan for what
will happen if that NIC fails.

> I configured service ip . Service IP is bound to ethernet network.
>
> I started hacmp service on one node..Vgs varied on and files mounted
> properly. I started hacmp service on second node. Hacmp services
> started on Second node and it got shutdown automatically after 1-2
> minutes. I need help to resolve this issue.

We need more information to discuss this. What did you see in
/tmp.hacmp.out?

#6 (**permalink**) 01-05-2008, 07:40 AM

MNI wrote:
> Hi
> I configured 2 AIX nodes and installed OS 5.3-ML 2 / HACMP 5.2 on
> each. I configured one ethernet interface on each node. I want to get
> rid of extra tcpip card usage by configuring 2 diskhb interface on each
> node. Is it OK ?
>
> I configured service ip . Service IP is bound to ethernet network.
>
> I started hacmp service on one node..Vgs varied on and files mounted
> properly. I started hacmp service on second node. Hacmp services
> started on Second node and it got shutdown automatically after 1-2
> minutes. I need help to resolve this issue.
>

It's a supported configuration, but you need to be using IP takeover by
alias, not by mac address. And I think that in this scenario you should
use some IP address for pinging.
--
BC

#7 (**permalink**) 01-05-2008, 07:40 AM

MNI wrote:
> HI
> Thanks all for your suggestion..I was so for using 3 tcpip [2 for
> network heartbeats] cards on each node. We are running short of NICs
> for a new 2 node cluster setup, thats why I want to give a try.
>
> Thanks for all your help.

As far as i can remember there is no need for NICs used just for
heardbeat. AFAIK hacmp uses every correct configured NIC ( in the sense
of IP ) for heartbeat. So with 3 Nic you have got almost 1 spare. ;-) (
Using IP takeover )

HACMP need NICs to communicate which each cluster member. Since HAMCP
in conjunction with hardware shell provide no single point of failure
you need at least 2 NICs. Its no problem to put any service ip alias to
any nic.

hth
Hajo

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

SEO

Disk HeartBeat ...Issue