VMware Cloud Community
nielsenmichael
Contributor
Contributor
Jump to solution

Problems with vmotion on Esxi

Hallo.

I have problem with vmotion migration between hosts.

It gives me the following error:

A generel system error occured: migration failed while copying data

and fails at 10%

HW configuration:

M1000e Blade enclosure

5*M610 blade server with 1 extra netcard for fabric b

2*I/O switch modules for Fabric A

2*I/O switch modules for Fabric B

There was not economy for 6 i/o modules so we have a configuration with only 4 switch i/o modules

So FabricB is Dedicated for ISCSI trafiic

Fabric A is then used for VM, Management trafiic and Vmotion

vmnic 0+1(Fabric A) is setup in a nic team on the virtual switch in a active standby configuration.

Vmnic0 active and Vmnic1 standby

On the vmkernel port on the same virtual switch i have changes the active standby order to vmnic 0 standby and vmnic1 active (I have also tried using the same active/standby order for the vmkernel as the virtuel switch setting but same error occurs).

So VM/Management is primary run on Vmnic0 and Vmotion traffic on vmnic1 primary

Vswitch0 on host1:

VM Network VLAN 168

Vmotion VLAN 168 ip:192.168.168.201

Management Network VLAN168 ip: 192.168.168.145

Vswitch0 on host2:

VM Network VLAN 168

Vmotion VLAN 168 ip:192.168.168.202

Management Network VLAN168 ip: 192.168.168.146

Tried the following without succces:

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=101315...

Offline migration works fine.

And it seems to be a periodic problem because last night it worked with an online vmotion but today same error again.

Any good suggestion on this issue?

Reply
0 Kudos
1 Solution

Accepted Solutions
FranckRookie
Leadership
Leadership
Jump to solution

Hi Michael,

VMotion does not care about storage network (as long as the datastore is reachable by both hosts). Only vmkernel ports with IP 201 and 202 are involved.

Check that only those two vmkernel ports are VMotion enabled. Are vmnic0 from both servers connected to the same switch? Try to catch the error message in /var/log/messages .

Good luck

Franck

View solution in original post

Reply
0 Kudos
7 Replies
illvilja
Hot Shot
Hot Shot
Jump to solution

Hi,

Can you verify with vmkping, too see if your VMotion interfaces can reach each other;

# vmkping 192.168.168.202

--

# vmkping 192.168.168.201




VCP3/VCP4/VTSP/VSP/MCTS: Virtualization

Reply
0 Kudos
nielsenmichael
Contributor
Contributor
Jump to solution

Ping works from both sides.

Reply
0 Kudos
illvilja
Hot Shot
Hot Shot
Jump to solution

Hi again,

Please check this link; http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=101315...




VCP3/VCP4/VTSP/VSP/MCTS: Virtualization

Reply
0 Kudos
FranckRookie
Leadership
Leadership
Jump to solution

Hi,

As you share your pNics between VMotion, Management network and VM network, it could be a network overload issue. Have a look at network usage when the problem occurs, especially because it works at night. You can also have a look at your physical switches too.

You should isolate your VMotion network!

Regards

Franck

Reply
0 Kudos
nielsenmichael
Contributor
Contributor
Jump to solution

It could not be performance problems because there is only 3 server running at the moment in a test setup and there is no activity on these server because the environment is not in production yet.

I have tried/tested everything mentioned in the links below and it is still now working.

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=101315...

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=100373...

Last thing mentionend in the article is that it could be releate to san misconfiguration. I cant fint any error in the san setup so i wil just list my setup below maybe someone can telle me if it is wrong.

esx01:

VMKernel port vmnic2 - IP 192.168.249.11 (Fabric A)

VMkernel port vmnic3 - IP 192.168.250.11 (Fabric B)

Under storage -> Properties -> Manage paths -> Set to Round robin (Vmware)

esx02:

VMKernel port vmnic2 - IP 192.168.249.12 (Fabric A)

VMkernel port vmnic3 - IP 192.168.250.12 (Fabric B)

Under storage -> Properties -> Manage paths -> Set to Round robin (Vmware)

Any good ideas.

Best regards

Michael

Reply
0 Kudos
FranckRookie
Leadership
Leadership
Jump to solution

Hi Michael,

VMotion does not care about storage network (as long as the datastore is reachable by both hosts). Only vmkernel ports with IP 201 and 202 are involved.

Check that only those two vmkernel ports are VMotion enabled. Are vmnic0 from both servers connected to the same switch? Try to catch the error message in /var/log/messages .

Good luck

Franck

Reply
0 Kudos
nielsenmichael
Contributor
Contributor
Jump to solution

Thank you for the assistance.

Checken the logs as FranckRookie recommended put me into the right direction.

It was cause by a IP adress conflict on the VMKernet network.

Now it works great

Reply
0 Kudos