have dell 1955 blades with (2) cisco 3030 switches fro ethernet. one fiber from each switch goes to the core. switch a = vmnic0 and switch b = vmnic1. when i disconnect fiber from switch a everything goes down and it isnt failing over to the other nic. in vc on all host i have the physical adapters teamed.
By default, ESX relies on the LINK state of the pNIC to detect failure. With the blades, the ESX host's link is still good (it is still connected to the switch), so ESX doesn't know to switch. Your best option is to create a link between the two blade switches so they can route between each other. That way, if there is a link failure, the blade switch will detect and correct, rather than relying on the vSwitch.
Ken Cline
Technical Director, Virtualization
VMware Communities User Moderator
whats the best way to create the link
Depends on the switches and the blade enclosure. I don't know about the Dell blades, but on the HP blades, there are two switch ports that serve as an interconnect between the two switches. You would typically trunk all VLANs between the two switches and they'll "be happy"
Ken Cline
Technical Director, Virtualization
VMware Communities User Moderator
In IBM blades you can set the switch to failover when external link fails.
In that way your ESX will see the loss of connection.
This is the most simple and working way.
If you like to check your ESX Nic teaming simply pull out the blade switch...
I recommend to look for switch fail options.
I'll try to examine this switch properties and assist you.
Regards,
David
i was able to fix it with beacon probing
dell blades 1955 cisco 3030 switches