vCenter alerts when scanning HBAs

[pironet]

Hi There,

I'm running vCenter 2.5U3/ESX3.5U3 with the latest patches for my ESX hosts.

I have noticed that I'm receiving false alerts when scanning my HBAs for new storage devices. The host appears with a red triangle in vCenter though it is still pingable on both SC's and all its guests are up and running...

Message is:

Target: esx502

Old Status: Green

New Status: Red

Current value: Host connection state - (State = Not responding)

Am I doing something wrong with the rescan ? Is it a bug ? Maybe vCenter mgmt agent is to sensitiv and I should increase a timed-out value ?

Thx for your input and regards,

Didier

[Troy_Clavell]

alarms are weird sometimes. I could be as simple as disabling the alarm at the Hosts & Clusters level, then enabling it again. Or you can restart the vpx agent on the Host(s) in question from the service console by issuing the command service vmware-vpxa restart.

Also, there alarm if in an HA cluster can mean there is something wrong with your configuration.

[pironet]

Hi Troy,

True that I'm running HA/DRS as well. vCenter doesn't say the HA agent has a problem and is disable...

The vCenter agent subset can't 'ping' the host whilst it's scanning HBAs and thus returns an alert and sends out an email.

Is there any settings I can modify in the vpxd.cfg for example ?

Thx,

Didier

[Troy_Clavell]

I'm not 100% this will fix your issue, but you may want to increase the Timeout Settings. Go to the adminstration field, drop down to VirtualCenter Management Server Configuration--Timeout Settings

[pironet]

Sorry but it doesn't work... This setting is between the vCenter Server and vCenter client... My issue is between vCenter and hosts.

[Troy_Clavell]

check to ensure this is not a name resolution issue. Can you ping your ESX hosts FQDN and shortname from your vCenter Server?

[pironet]

Names resolution are fine...

I have noticed that if I ping continuously the SC whilst I do a HBA scan, I have some timed-outs and at the same time vCenter has the host disconnected.

What's the impact of a HBA scan over a host ?

Note that the host boots up fron SAN, could that be the scan 'disconnects' the boot LUN for some times ??

Thx,

[pironet]

Problem fixed and understood. Name resolution is fine.

Edge network device (Cisco 3020s) are not...

We use active/active vmnics, and therefore set the vSwitch load balance method to "loadbalance_ip" BUT edge Cisco switches are configured with proprietary etherchannel trunks and it doesn't work properly with ESX hosts.

We set it back to "loadbalance_srcip" which works but not the best practice and we will reconfigure properly the Cisco's at a later time.

Rgds,

Didier