Problem with "regonfigure for HA" after Virtual Center upgrade.

[sadlyzer]

Hello,

I Have a problem that hit me after I upgrade my Virtual Center.

Old version: 2.0.1

Build: 33643

New version: 2.0.2

Buid: 62327

Upgrade process went went, but after I run Virtual Center Client I soo that all Nodes were Disconnected. I tryd to connect but with no success.

OK.

I copy vpx-upgrade-esx-6-linux-62327 to nodes and run "sh vpx-upgrade-esx-6-linux-62327". I got no error. After that I connect nodes in Virtual Center and this time connection progress went fine but when I try to reconfigure HA then the status counts to 95% and after that gives a error: An error occurred during configuration of the HA Agent on the host.

Is there some normal solution to that?

Raul

[jhanekom]

The cause of your first problem was probably a missing /tmp/vmware-root folder (plenty of references to this in the forums.)

The cause of your second problem is a bit more difficult to troubleshoot. I've encountered cases where it seems I need to reconfigure the "primary" HA node first, though it's fairly difficult to tell which that is. The easiest way is usually to make your way through all the servers, doing "Reconfigure for HA" one at a time until one of them succeeds. The rest will then usually succeed after that.

A bit too unscientific for my liking, but it's worked for me so far.

(The other most common cause of HA breaking is name resolution, but since it presumably worked before in your environment, that shouldn't be a problem now.)

[sadlyzer]

Hi,

Thank you for answering.

1. I checked - /tmp/vmware-root directory exist in all nodes.

2. As I write this letter I also dowload VC 2.5 iso. I try with that. It seems to me that the problem is related with network. The problem is that when I connect node then it communicate directly (using name or IP) but when I try to reconfigure HA then it try to communicate through gateway (that by the way doesent exist)....this solution worked fine with old version of VC but I guess not with new version. (I use it because I want to hold vmware network isolated)

I guess I have here two choices:

1) I put nodes to existing subnet (with gateway)

2) I configure a vlans and use different vlan with vmware.

As I see it the nr 2 is the best option, but there are also some problems:

1) I have 4 nodes and each node has 4 NIC. I use teaming mode. So it means that I need to use same ports in switch witch means that I need to use tagged vlans witch means that I need to configure NIC's to support vlans. Write now I do not know how to to that in ESX environment.

Any ideas?

Raul

[zbenga]

usually the console port needs to be on the same subnet as vc, makes things easy

if any dns files changed on the hosts HA will play up, check /etc/hosts and make sure they are the same, eg. ip host.domain.xxx host

[admin]

Try disabling HA on the cluster. Wait for the unconfigure tasks to complete on all the hosts. Then re-enable HA on the cluster.

[blevinsa]

Check the network mask of the service console if this is happening on only one server.

[depping]

so what address do you use for isolation ? did you set this via advanced options? cause you need an isolation address which HA can check if it thinks it's isolated. HA will check for a pingable isolation address when you enable it.

if you've set up an iso adress than you can disable the usage of the GW by setting the following in the advanced options section of HA on your cluster:

das.usedefaultisolationaddress - false

Duncan

My virtualisation blog:

If you find this information useful, please award points for "correct" or "helpful".

[mike_laspina]

Ditto on the disable HA at the cluster level and wait for it to fully uninstall before re-enabling. Worked for my lab testing.