HA fails to enable because service console has no redundancy?

[bateslabel]

I am resetting up our vmware. It was initially done by a consulting company, but after many crashes and lack of updates, I decided to rebuild it. I have updated all the hosts and Vcenter to 4.1. Here is my network setup. Each hosts has two nics. One nic is assigned to the virtual machine network and the other to the managment network. I have two vmkernals on each virtual management switch. One for vmotion and the other for management. I am new to configuring a setup by myself from scratch. Everything has went smoothly until now. I have worked from vmware pdfs and notes from the compnay to get this far. How do I configure the service console for redundancy? Isn't that the vmkernal port for management traffic? Clearly I am missing something. I have spent hours trying to search the issue, have come across all kinds of stuff, but no clear and definate explanantion. One more thing I don't think this is an issue but it may be. Do the swicthes on each host have to have the vmkernals be on identical vmkernal ports. On one host the vmkernal port is 1 for vmotion traffic, but its vmkernal port 2 on the other host. Correct ip info, identical switch names, and I have the hosts configure to use the seperate backend management network. I used dns, two entries one to resolve the host on the dedicated backend network for vcenter and another dns entry to resolve the host from the front end network. What am I doing wrong?

[idle-jam]

All you need is more than 1 vmnic for that error message to go off. Also you can enable HA even with 1 vmnic. Is the ESX hostname resolvable between all the hosts and vcenter? you will need those for the vmware ha to work.

[ThompsG]

Hi,

Just so you are aware having redundancy on the Service Console/Management network is not a requirement to enable HA so it would be my guess that something else is wrong. Have you confirmed that you have DNS resolution between hosts as this is a requirement and without it HA will not enable? Perhaps if you could post a screenshot of the vSwich layouts we might be able to see what the issue is. Reason for asking this is you have said this:

I used dns, two entries one to resolve the host on the dedicated backend network for vcenter and another dns entry to resolve the host from the front end network.

Which sems to indicate you have two dns entrys for the Service Console.

Now I know you are probably already thinking this but I would definitely be looking to add additional network adapters to your hosts as two is pushing it to make a decent design.

Anywho on to the matter at hand - given that you only have two network adapters in the ESX servers and it is nice to have a level of redundancy I would probably configure with one vSwitch with two nics.  On this one vSwitch I would have multiple portgroups and assuming your nics are vmnic0 and vmnic1 I would configured as follows:

• Service Console/Management network - vmnic0 (act) / vmnic1 (standby)
  
• vmotion - vmnic0 (act) / vmnic1 (standby)
  
• virtual machine network - vmnic0 (standby) / vmnic1 (act)
  

Now normally you would want to keep your network traffic seperate especially the vmotion traffic but if you want to have network redundancy for the Service Console then you really have little choice.

Kind regards,

Glen

[depping]

This tech paper explains the importance of Service Console Redundancy and shows an example:

http://www.vmware.com/files/pdf/techpaper/VMW-Server-WP-BestPractices.pdf

Duncan (VCDX)

Available now on Amazon: vSphere 4.1 HA and DRS technical deepdive

[VMrOxALL]

Hi,

I agree with ThompsG that DNS could be the issue. You can try pinging the hostname from each host, but a true test is to Vmotion a VM between each host.  If that works then you rulled out DNS, vSitch names, etc.

[bateslabel]

It dns issue. I am working out the configuration.

[bateslabel]

I have it worked out

[aparis]

hey

I reccomend to fix DNS problems in a HA cluster to modify manualy the hosts file in all the cluster ESX.

Use DNS for name resolution rather than the error-prone method of manually editing the local

/etc/hosts file on ESX hosts. If you do edit /etc/hosts, you must include both long and short names. VMware HA modifies the /etc/hosts file if the information in it is wrong or needs to be added.

Check this document to because there describes how to customise HA cluster to make it response as you like and no by default.

Check the page 26.

http://www.vmware.com/pdf/vsphere4/r40/vsp_40_availability.pdf

I hope this information helps you too configure better your HA cluster.

Bye

Andre Paris

Codic

andre.paris@codic.com.uy

skype: andre.paris