Hosts keep showing up as not responding in vCenter

[Legacy User]

I'm running vSphere Web Client  version 6.0 and it has three hosts in one cluster. 

I notice that two of my hosts keep going into a "not responding" state after about five mins. 

I can right click, reconnect and it's all good, then it goes into not responding again in, like I said, about five mins.

I did just recently disconnect and reconnect these hosts, but it's only two out of the three that's giving me grief. 

When these hosts go into not responding, they are still pingable and the VMs are still running fine, so it's not that they are down at all.

Any thoughts as to how to resolve this?

Thanks

Cliff

[golddiggie]

Check the Host Management network configuration. You could also try restarting the host management from the hosts (connect to the host via KVM/iDRAC/iLO/etc logging in directly). I'm sure someone will post up about logs you can pull, but I've not dived too deep into those items (yet at least).

[Vijay2027]

What is the vCenter version and is it installed on windows or appliance?

[SP2911]

Hello,

Could you please upload any one of the ESXI host logs and the Vcenter log so that we can verify the same

meanwhile, Do restart the Following services

/etc/init.d/hostd restart

/etc/init.d/vpxa restart

Do monitor the status

[SP2911]

Hello,

Could you please upload any one of the ESXI host logs and the Vcenter log so that we can verify the same

meanwhile, Do restart the Following services

/etc/init.d/hostd restart

/etc/init.d/vpxa restart

Do monitor the status

[IRIX201110141]

When a host becomes to the "not connect" state in 1-2min after you reconnect or added it to the vcenter its based on the following behaviour. The connecting starts with a request to 443 or so  get certificate stuff and enabling and configuring of the vpxa service on the host. Now vCenter waits for a answer from the vpxa and when there nothing comes back a timeout kicked in an the host drops again out of vcenter.

So as already suggest try to restart the mgmt service on the host. But iam not a fan for using the etc scripts directly. If you can connect by using ssh you can enter "dcui" and after login again try "troubleshooting options" -> "Restart mgmt service".  Of course you can use ILO/iDRAC also. While there you can take a look to the vpxa.log.

Regards,

Joerg

[Paul_Knight1]

When the hosts disconnect, are you still able to get to the console?  Are you using SAN disk arrays? An issue I've seen in the past, is a host floods its array with IOPs, and its mgmt services queues start to back up.  The host becomes unresponsive until the flood subsides, to the point where you couldn't even get a usable bumble-bee screen.