Updated from 3.5U1 to 3.5U2 and now get "Unable to automatically migrate xxxx"

[ardenen]

Hello,

We just recently migrated our two clusters to 3.5U2 from 3.5U1. We updated all servers to 3.5U2 and the VI to 2.5U2. Now all the servers are unable to automatically migrate VMs when we attempt to put them in maintenance mode -- we can manually migrate/vmotion. I found all this out the hard way with a 1am SAN maintenance/update that required a reboot of the ESX hypervisors (not fun manually vmotioning all our boxes).

I've searched the web and this forum with no luck... anyone else experience this issue and find a fix?

Thanks!

[lmonaco]

Have you verified if your DRS settings are still set to fully automated?

[ardenen]

Yes -- I found a few notes/threads here recommended disabling DRS so I tried that as well -- no luck.

Also -- once I've manually migrated all the machines to another server, entered maintenance mode, then exited maintenance mode they will automatically load balance the available servers and vmotion machines back to the serviced server.

[markvor]

what kind of nic's are in the vm's

any entroes in the logfiles of the esx servers ?

carpe diem

markus

[ardenen]

One cluster has quad port Broadcom NetXtreme II BCM5708 1000Base-T nics and the other cluster has dual-port Broadcom NetXtreme II BMC5708 1000Base-T and Intel PRO/1000 PT Low Profile Quad Port PCIe Gigabit Ethernet Controller.

I replicated the issue to see if there were any changes in the logs

vmkernel: nothing

hostd.log:

Hw info file: /etc/vmware/hostd/hwInfo.xml

Config target info loaded

CheckoutVdiVmxLicense: key esxVmx not found in license source

UpdateVdiVmxReservation CreateReservation failed, reporting to client is offline

I double checked and according to the VIC all servers have valid licensing and I double-checked I had the latest patches -- everything appears up-to-date.

[jnorred]

See my post a few minutes ago...

Try disabling HA to see if that allows you to put your hosts in maintenance mode. Started having this problem after my VC Update 2 upgrade.

Jason

[ardenen]

If you look above you will see I tried disabeling HA before I posted the question. Thank you for your suggestions though.

[jnorred]

DRS or HA? If you have already tried HA and are still having problems then thats a new one on me....

[ardenen]

Ooops -- just noticed I only mentioned DRS -- yes DRS and HA -- I have tried disabeling one then the other and I believe I tried with both disabled... no luck.

[Dandan712]

I've heard that there is still an issue with update 2 regarding automatic migrations.

It may be a good idea to remain Update 1 for a while...

But regarding your issue, I 'm not sure if this has an impact or not.

[jamesl73]

I'm also at Update2 and just noticed this same problem with one of my

clusters. However I have another cluster using the same VC that does

work. Working cluster is 10 nodes. Non-working cluster is 2 nodes.

[Dandan712]

Yes.

All this reinforce my opinion to keep on wokring with update1 and wait for a lter update. You can be nearly certain that any future update from vmware will be far more verified anf checked than this one, because of all the issues that were raised upon update2

[cseeliger]

With VirtualCenter 2.5 U3 it's possible to switch HA settings to "allow VMs to be powered on even if the violate availablity constraints" to avoid the problem.

see also:

Regards,

Christoph.

[ardenen]

This has been an option for a while. However VI 2.5.U3 does appear to fix many of the issues we were having with 2.5.U3 -- so far so good with our upgrade.

[geob]

We had the same issue, and after doing a vmkping from one host to another, the migrations starting working, and then continued to work.

[ardenen]

I'll have to remember that for next time... for now updating to 2.5U3 fixed our issues.

[Koutsi08]

A couple of days ago updated two ESXs to 3.5u3 and VC to 2.5u3. Had the issue of "unable to automatically migrate xxxx". I did these and finally Entering Maintenance Mode works with automated migrations:

1) Fixed and synchronized ESX clock times at the service console by Unix command 'date 11121610' (10 min over 4 pm 12th of November).

2) Rebooted ESXs.

3) Rebooted VC server.

4) At cluster settings put VMware HA option to "Allow VMs to be powered on even if they violate availability constraints".

5) Uninstalled test virtual machine's (test VM for the maintenance mode migration) VMware Tools ver 110xxx, rebooted VM and installed again the VMware Tools which now are of version build 123630 (same as ESX). VMware Tools hadn't told it requires update (as I would have expected from earlier behaviour).

I guess one or two of those measures are not needed. Probably the 3rd and 5th steps were the most important.