guson99
Contributor
Contributor
‎01-20-2011 10:58 AM

VM power off when cable red is crash HA

Hi, I'm new in this platform and I installed Vphere 4.1, my environment is: 
1 Storage 
3 host 
1 Vcenter with Datacenter created 
The problem is in the HA apparently, I configured HA and now test my environment before going to productive. 
I configured all vswitch with 2 nics, both to 1GB. 
Today I change the cable of one nic of the HOST 1 and the VM in host 1 continuous running OK, but when I reconnect the cable in the nic, the VMWARE POWER OFF my VM and restarted in the same HOST 1. 
Why occurs this if my second nic of host 1 never lost connection? 
is this correct or my vm never has power off? 
 
Sorry by my english is not very good. 
Regards.

Reply
0 Kudos
cSonnierEsxi
Contributor
Contributor
‎01-20-2011 02:23 PM

How are you secondary nics setup in your vSwitch? Standby, Active, etc.

Reply
0 Kudos
a_p_
Leadership
Leadership
‎01-20-2011 03:01 PM

Unless you configured something special, there may only be two reasons for this:

  1. the physical switch ports are not configured for spanning-tree portfast
  2. you did not increase the isolation response time - HA advanced settings -> das.failuredetectiontime = 30000 (30 sec)

The 30 sec are best practice and may be adjusted for your needs. Default setting is 15000 (15 sec).

André

Reply
0 Kudos
guson99
Contributor
Contributor
‎01-21-2011 10:39 AM

Hi, thanks for response. 
My 2 nics are in active mode in the vswitch. 
The 2 nics are connected to the same switch Cisco 3750, but I not configured nothing with spanning-tree portfast 
maybe this is a my problem. 
Now, if I have another switch and I connect 1 nic connect to the cisco 3750 (1) and the another nic 2 connect to the cisco 3750 (2), I need spanning-tree portfast? 
Do I have the same a problem?  
Sorry for my unknowledge

Regards.

Reply
0 Kudos
a_p_
Leadership
Leadership
‎01-21-2011 11:04 AM

You'll have to configure spanning-tree portfast on each switch port, no matter if it is on the same switch or another. Unless you set it, the spanning-tree protocol can take up to ~45 seconds to allow traffic on a port in case of an uplink failure. Since HA is triggered after 15 seconds (by default) you would always run into HA once an uplink fails.

André

Reply
0 Kudos
guson99
Contributor
Contributor
‎01-24-2011 05:58 AM

Hi Andre, thank you for your recommendation.

Regards.

Reply
0 Kudos