HA configuration error

[francescog]

Hi all,

I have a 11 nodes cluster. Shared storage iSCSI, thus every host has two Service Consoles for iSCSI connection.

On one of the host I get this error when I try to configure HA:

"hostipaddrsdiffer: cmd remove failed: ft_gethostbyname and hostname -i return different address: ... (here it lists the ip address of the SECOND Service Console)

When I issue the command /opt/LGTaam512/bin/ft_gethostbyname "hostname" I get the ip address of the Service Console 2.

Anybody knows how to get out of this hell?

Thank you in advance.

[Diego_Mora]

Remove the second service console port group. Move all virtual machine to another host. Disconnect that host and remove it from cluster. Adding again to cluster and recreate the second service console port group.

[francescog]

Thanks for the prompt reply.

Unfortunately I already tryed to do what you suggested.

No way. It seems that HA has its own "hosts file" and doesn't give a damn of the DNS and the /etc/hosts file.

Do you know if there'a a way to "reset" the information of the HA agent?

Anyway, HA s#cks!!!

[Diego_Mora]

Has you remove the host from VC inventory too? When you remove a host from VC Inventory it initiate a procces that uninstall LGTO (legato packages for HA) and VPXA (VC agent) from the host.

And when you add a host to VC inventory it install those packages.

[francescog]

Yes Diego, I did it.

These are the step I followed:

1) removed host from the cluster

2) removed Service Console 2

3) rebooted the host

4) removed the host from the VC

5)re-added the host to the VC (at this point it creates the LGTOaam512 directory)

6) BEFORE adding the host to the cluster I re-issued the command /opt/LGTOaam512/bin/ft_gethostbyname and I'm still getting the wrong IP address!!!

[dinny]

Hiya,

I would have a look in /etc/FT_HOSTS

That seems to be the file that HA actually uses for it's host info.

However that should be created OK from other host info when you configure the HA service.

By the way lots of duplicate entries per node seem to be normal....

I usually stop and start HA on all the nodes in my cluster when I add a new node - that way I know that all the existing nodes know about the new one.

If one node does have dodgy info in the FT_HOSTS and you need to clear it totally - perhaps remove it, then re-add it to the HA cluster?

Hope this helps...

Dinny

[Diego_Mora]

What is the result of hostname -i command?

In the /etc/sysconfig/network-scripts directory you will see a file named (or similar to) ifcfg-vswif0. How many of this files can you see?

[chrisknoche]

Not sure if you still are having issues with this but remove the duplicate IPs from the /etc/host file on the esx server.

Doing so fixed my problem.

[DeanatHP]

Something else to check is to make sure your DNS server is serving the same IP address for your ESX Server host as what the host is configured for.

Dean

[gbowers]

I followed this proceedure, no reboot required -

Place host in maintenance mode

Add new service console (different IP) in GUI

FROM a console log into ESX host

run esxfig-vswif –l (list all vswif)

run Esxcfg-vswif -d vswif0 (your old service console)

run service mgmt-vmware restart

run service network restart

vi /etc/sysconfig/network -> change vswif0 to vswif1

vi /etc/sysconfig/network-scripts/ifcfg-vswif1 -> modify IP address to old IP

service network restart

service mgmt-vmware resta

[PhilipAvis]

I did this and it worked.

You have to ensure that the correct DNS entries are in place.

Then you have to go into the /etc/hosts file and put in ALL the cluster members

ie

x.x.x.x cluster01 cluster01.domain.com

x.x.x.x cluster02 cluster02.domain.com

do it on your virtual centre as well in the local hosts file and everything should be OK. No restarts, no maintenance mode etc.

I had to take HA off and then re-add it after I had made the host changes.

confirm all is right with

hostname

hostname -i

hostname -s

Vish

Message was edited by:

Vish

[Tek_King]

Thanks Dinny,

Your solution resolved the issue at the top of the thread, sorry it was the last solution on this thread I tried.

Craig.

[dinny]

Hiya,

Glad to hear it.

You can award the 10 points and mark the question as correct by clicking on the "correct" button shown by my reply.

Cheers

Dinny

[Trigsy]

FT_HOSTS cleaned out, worked for me. Thx

p.s. Thanks for making me use VI again

[gotts]

My host file was set up as xx.xx.xx.xx hostname.domain.com hostname. Upon viewing your example I changed my host file to xx.xx.xx.xx hostname hostname.domain.com and then rebooted my esx server. Upon completing the reboot HA was porperly configured. I am not sure that this really made a difference because it had worked previously in the format I had but it works again and for now that is all that matters.

Thanks to all for your help!