VMware Cloud Community
stkj_ing
Contributor
Contributor
Jump to solution

losing link on the management makes all VMs shutdown!?

Hi!

I have had this problem sometime now. I have two nics in my ESX servers, each nic is connected to a sepparate switch. I then use VLANs to set upp management and vm-networks.

My management network have both nics connected (nic0 as active and nic1 as standby). This worked fine in tests in the 3.0.2 version but I now have upgrade to 3.5i and yesterday

when i rebooted the primary switch all my VMs where shutdown!

Is there any best practice in this, becouse I think this is a BIG problem with ESX right now.

Reply
0 Kudos
1 Solution

Accepted Solutions
Chris_S_UK
Expert
Expert
Jump to solution

I am presuming your hosts are HA enabled......HA has obviously detected a host isolation. The default behaviour in this scenario is to power off the VMs.

Possible courses of action.

1. Check your pswitch and vswitch config to ensure that, if one pnic goes disconnected, the other one picks up the link properly

2. Increase the timeout using the das.failuredetectiontime setting in the HA settings (suggest to 60000 msecs)

3. Change the VMs' settings in HA to not power down. Note however that this setting will stop them being powered up on another host if there is a protracted network issue with a host (i.e. such that even with an increased isolation time, HA detects host isolation)

Chris

View solution in original post

Reply
0 Kudos
2 Replies
Chris_S_UK
Expert
Expert
Jump to solution

I am presuming your hosts are HA enabled......HA has obviously detected a host isolation. The default behaviour in this scenario is to power off the VMs.

Possible courses of action.

1. Check your pswitch and vswitch config to ensure that, if one pnic goes disconnected, the other one picks up the link properly

2. Increase the timeout using the das.failuredetectiontime setting in the HA settings (suggest to 60000 msecs)

3. Change the VMs' settings in HA to not power down. Note however that this setting will stop them being powered up on another host if there is a protracted network issue with a host (i.e. such that even with an increased isolation time, HA detects host isolation)

Chris

Reply
0 Kudos
stkj_ing
Contributor
Contributor
Jump to solution

Thanks for the fast answer!

1. I have gone through my switches and found that they where configured a little different. (Is there a way to configure networking on the cluster so that all machines have the same network settings?

2. sounds a little to advanced

3. That must have been it.. the cluster setting was to power off VMs when isolated...

Reply
0 Kudos