Agent Unreachable after logoff of linked clone, OS has 169.254 apipa address

[WheatonCollege]

Hi everyone,

We've noticed some Win7 linked clones remain in the "Agent Unreachable" state for a while before getting refreshed. We think we narrowed it down to these clones ending up with a 169.254 (apipa) address instead of a DHCP address on our network. Our lease times are only an hour so I believe the clones in this state try to get a DHCP address 30 minutes later and it fixes itself. If we manually go into a clone exhibiting this behavior and try and get a valid DHCP address using ipconfig it gets one and the system gets refreshed.

Here are some things we've researched already.

1. The systems only have one virtual NIC and it is a VMXNET3 NIC

2. Within "Edit Settings" for a VM it is showing the NIC as "Connected" and is set to "Connect at Power on"

3. The VMware View agent service is running within the VM

4. The VM is listed as running the latest, current, VMware Tools inside of vCenter

5. This has happened on different ESXi hosts so it doesn't seem limited to one host.

6. We checked the DHCP scope on the DHCP server, of the 900+ addresses only 8% are being used so there are plenty of addresses available

7. As mentioned earlier, if you go into the VM's console and do an ipconfig /release then /renew it gets a valid address and completes the refresh process

8. Within the virtual network switch the number of ports (120 per switch) is well above the number of VM's on the switch (at the most about 15 VMs).

9. We checked the DHCP server logs and it doesn't look like the VM's are making a request for a new address on it's own (at least not during the logoff time frame).

The View environment is version 6.1.1 build 2769403

Any suggestions would be greatly appreciated 🙂

[PCTechStream]

That might happen when:

- The VM fails to join the Windows Active Directory domain

- View Connection Server can't communication with the View Agent on a the VM

- When the VM is not getting the DHCP address when provisioned

- Try ipconfig /renew on the VM getting 169.254 & see if you get a valid IP

- Try to Pings a specified host until stopped. To see statistics  use -t

- Try doing ipconfig /release & ipconfig /flushdns on the master VM before shutting it down for new snapshot

- Check your DNS & DHCP scope size & IP lease expiration. Maybe it's a DHCP problem & No a Horizon View issue

- Check the user account & OU you use to recompose the pools, sometimes VDI is not able to update the DNS records and therefore failed to receive the DHCP addresses. Finally check your antivirus.  

Raul.

VMware VDI Administrator

http://ITCloudStream.com/

[WheatonCollege]

Hi Raul, thanks for the suggestions... In response to your comments I've included my responses in red

- View Connection Server can't communication with the View Agent on a the VM - This is true because the VDI desktop with the view agent doesn't have a valid IP address

- When the VM is not getting the DHCP address when provisioned - This is true

- Try ipconfig /renew on the VM getting 169.254 & see if you get a valid IP - This does fix the issue

- Try to Pings a specified host until stopped. To see statistics  use -t - Once I manually release/renew the IP and it goes from a 169.254 address to a valid address everything ping fines and the VDI desktop refreshes itself

- Try doing ipconfig /release & ipconfig /flushdns on the master VM before shutting it down for new snapshot - We do am ipconfig /release on the master images but not an ipconfig  /flushdns, we will add that to our prep batch scripts.

- Check your DNS & DHCP scope size & IP lease expiration. Maybe it's a DHCP problem & No a Horizon View issue - The scopes are fine, we are only using 8% of our available 900+ IPs in that subnet

- Check the user account & OU you use to recompose the pools, sometimes VDI is not able to update the DNS records and therefore failed to receive the DHCP addresses. Finally check your antivirus.  - Our DHCP server takes care of the dynamic DNS update on the client's behalf. Once I manually release and renew the DHCP settings the DHCP server registers the VDI system's IP in DNS

[PCTechStream]

Take a look at this thread

LINK: desktops linger in "Agent Unreachable" folder for a long time, apipa address might be a clue

You might have connectivity issues when the VMs can't get a valid IP address automatically. Make sure there is only one logical route between the DHCP server and the remote subnet clients.

[TechMassey]

This is a really detailed post, thanks for those details. Based on what is listed, I would first want to rule out the IP helper for the VLAN you are using. Just to verify that deprecated DHCP servers are not listed in the IP helper. My other question would be if this is a recent or long standing issue?

[kevinpower]

Hello,

Two weeks ago we got the same problem, after a time zone change ( 1 our forward, summertime ) our VDI clients didn't request any ip address from the dhcp server after a power on action.

What we did;

-Power on the master vm

-Checked if the time zone was correctly

-Shutdown, and make a snapshot

-Recompose the pool

Good luck