vCenter

Of our 4 ESX servers in a HA cluster, I have one with HA agent failures. When I try to configure HA it will time out. The vpxa.log wouldn't tell me anything useful. Reconfigure isn't working either. Removing the agent when I disable HA on the cluster will also time-out. Rebooting the server didn't help.

Can I remove the agent manually from the host in that way that enabling HA will configure the agent the correct way?

The HA agents write to /opt/LGTOaam512/log/aam_config_util_addnode.log during configuration. Check that to see if it indicates the problem.

It's probably a DNS issue, check that the problem host can ping itself and the other ESX hosts in the cluster by hostname only[/b]. The HA agent (aka Legato AAM) only uses shortname to contact the other hosts.

If you can't ping the other hosts, add entries to /etc/hosts on the server for each ESX host in the cluster. Also make sure you have put your search domain in /etc/resolve.conf.

You should have a line like:

search yourdomain.com[/b]

at the top of the file.

If you still can't get the agent working and the DNS is defnitely working as it should, put the host into maintenence mode then remove and re-add it to the cluster. 90% of the time HA problems are DNS, the other 9% of the time removing and re-adding the host from the cluster sorts the problem.

I checked the aam_config, but it hasn't been written for a long time. Probably because "reconfigure" isn't doing anything but timing out.

DNS resolving is oké, resolve.conf is correct. All other ESX server can be pinged with and without FQDN.

I changed the server to maint mode, removed from cluster, rebooted (never know) and moved it into the HA cluster again. Inmediately the error: "an error ocurred during the configuration of the HA agent on the host" appeared. util_addnode has not been touched. listnodes.log had the following entry:

CMD: /opt/LGTOaam512/bin/ft_gethostbyname |grep FAILED

Is there a way to remove completely the AAM software from the host and reinstall it? Otherwise I will just choose to reinstall ESX

tnxs.

Paul

Also:

in VC it gave the error:

:cmd enable firewall failed

Hmm, could be that the firewall config via the VI client isn't working then and it isn't opening the port required for the HA agent.

On the command line type esxcfg-firewall -q[/b] and verify that AAMClient is listed as an enabled service. If it isn't type esxcfg-firewall -e AAMClient[/b].

You could check the content of the /etc/hotst file and see if the ip adresses and dns names match with the SC of your esx server.

I had a simmilair issue, and it came down to this:

when i installed esx i assigned a fictive IP adres to the SC, that IP gets in de /etc/hosts file.

if you change the IP adres of the SC the /etc/hosts file does NOT get updated, and configuring HA fails

we all know that HA is highly dependant on DNS, but host files take precedence as we know, so be sure that it matches

IP adress \[tab] FQDN \[tab] hostname

tnxs for the tip. I double checked ip settings and DNS for all hosts anyway. We don''t want HA if it's not so HA, don't we?

I think you touched correctly there.

\[root@iscbse0002 root]# esxcfg-firewall -q

modprobe: Can't locate module ip_tables

iptables v1.2.8: can't initialize iptables table `filter': iptables who? (do you need to insmod?)

Perhaps iptables or your kernel needs to be upgraded.

Incoming and outgoing ports blocked by default.

Enabled services: CIMSLP CIMHttpsServer vpxHeartbeats AAMClient LicenseClient sshServer CIMHttpServer

Opened ports:

\[root@iscbse0002 root]#

Well, it listens to ssh, because I connect with putty, but this sounds very wrong.

Yep, sounds like your installation of iptables is knackered. You can try and re-install it from an RPM but if the system isn't live or you have the capacity to migrate the VMs off the host I'd just go for a rebuild.

Good thing that reinstalling is such a simple thing. Maybe even faster then reinstalling the stack.

tnxs

I had this happen to me today. Turns out I had a typo on my hostname for the box. Just wanted to post this for something else to check if you are having problems.