Yellow Bricks

by Duncan Epping

ha

HA advanced options

· ·

PeterB posted a topic about all known HA advanced options on the Dutch VMUG. If anyone has more options please post ’em and let’s try to make this list as long as possible:

  • das.failuredetectiontime – Amount of milliseconds, timeout time for isolation response action
  • das.isolationaddress[0-9] – IP address the ESX hosts uses validating if it is isolated from the network or not. It will use the default gateway by default. With this setting additional addresses can be configured for use.
  • das.poweroffonisolation – Values are False or True, this is for setting the isolation response. Default a VM will be powered off.
  • das.vmMemoryMinMB – Higher values will reserve more space for failovers.
  • das.vmCpuMinMHz – Higher values will reserve more space for failovers.
  • das.defaultfailoverhost – Value is a hostname, this host will be the primary failover host.

ESX 3.5 and weird DRS/HA behaviour…

· ·

During the last couple of ESX 3.5 and VC 2.5 implementations I encountered some weird DRS/HA behavior:

  1. During the reboot of a VM ESX decided to vmotion the VM, as you can imagine this took a lot longer than normal due to the memory being excessively active.
  2. When the DRS Affinity Rules in a two cluster node is set to “seperate” two specific virtual machines VMware has a hard time relocating a VM when a host goes into “maintenance mode”. I can imagine why this happens but it’s definitely not something that’s suppose to happen in my opinion. It’s a Admin initiated manual action it should overrule all affinity settings.

So let me hear which weird behavior you noticed with ESX 3.5 / VC 2.5….

Xen HA clustering

· ·

The company I work for, Ictivity, recently started testing with Xen(not Xensource!) and HA clustering. There were two case studies, one based on Suse and one based on Red Hat. Both definitely have it’s own advantages and disadvantages but it seemed promising to me. Today I noticed a blog on ONLamp.com about how to set this up. It’s a good read and gives you insight information. You can do this with open source only software so it will not get any cheaper than this. If this solution is suitable for an enterprise environment is definitely a point of discussion, but it is good to see there are more alternatives coming our way.

The idea of using virtual machines to build high available clusters is not new. Some software companies claim that virtualization is the answer to your HA problems, off course that’s not true. Yes, you can reduce downtime by migrating virtual machines to another physical machine for maintenance purposes or when you think hardware is about to fail, but if an application crashes you still need to make sure another application instance takes over the service. And by the time your hardware fails, it’s usually already too late to initiate the migration. So, for each and every application you still need to look at whether you want to have it constantly available, if you can afford the application to be down for some time, or if your users won’t mind having to relogin when one server fails.

New HA error messages

· ·

Just noticed the new errors HA produces at the console when a host is isolated in ESX 3.5:

Node isolated from network. Either repair network or override agent startup using option -noiso

And when the host returns from isolation:

Node no longer isolated, will restart

This makes troubleshooting a lot easier! Thanks VMware 🙂

Support on ESX 3i and HA

· ·

I never noticed this, but during the configuration of HA in my testlab I received some weird errors. I just could not enable it on the ESX Server that ran ESX 3i, and I’m pretty sure DNS works fine. Today I received an email about 3i and support on HA, which states:

Support for VMware High Availability (HA) is experimental and expected to be fully supported in an upcoming patch or update.

You can read more about it in this PDF about 3i and HA. But is it me or is almost every single interesting new feature experimental. I hope that VMware will support all the new features when the patches/updates are released