Troubleshooting virtual machine live migration


Troubleshooting virtual machine live migration
Prev	Appendix A. Troubleshooting	Next

A VCS cluster is formed between virtual machines (VMs) and one of the virtual machines is migrated from one host to another host. During a virtual machine migration, if the VM takes more than 16 seconds to migrate to the target node, one of the VMs panics. In this case, 16 seconds is the default value of the LLT peerinact parameter. You can increase the peerinact value to allow sufficient time for the VM to migrate. You can adjust this time based on the environment in which you initiate the VM migration.

To avoid false failovers for virtual machine migration, you can change the peerinact value using the following methods:

Set the peerinact value dynamically using lltconfig command:
```
# lltconfig -T peerinact:value
```
Set the peerinact value in the /etc/llttab file to make the value persistent across reboots.

To set the peerinact value dynamically using lltconfig command

Determine how long the migrating node is unresponsive in your environment.
If that time is less than the default LLT peer inactive timeout of 16 seconds, VCS operates normally.
If not, increase the peer inactive timeout to an appropriate value on all the nodes in the cluster before beginning the migration.

For example, to set the LLT peerinact timeout to 20 seconds, use the following command:
```
# lltconfig -T peerinact:2000
```
The value of the peerinact command is in .01 seconds.

Verify that peerinact has been set to 20 seconds:

# lltconfig -T query

Current LLT timer values (.01 sec units):
 heartbeat   = 50
 heartbeatlo = 100
 peertrouble = 200
 peerinact   = 2000
 oos         = 10
 retrans     = 10
 service     = 100
 arp         = 30000
 arpreq         = 3000
 Current LLT flow control values (in packets):
 lowwater  = 40

Repeat steps 2 to 3 on other cluster nodes.
Reset the value back to the default peerinact value using the lltconfig command after the migration is complete.

To make the LLT peerinact value persistent across reboots:

Append the following line at the end of /etc/llttab file to set the LT peerinact value to 20 seconds:

set-timer peerinact:2000

After appending the above line, /etc/llttab file should appear similar to the following:

# cat /etc/llttab
set-node sys1
set-cluster 1234
link eth2 eth-00:15:17:48:b5:80 - ether - -
link eth3 eth-00:15:17:48:b5:81 - ether - -
set-timer peerinact:2000

For more information on VCS commands, see the Cluster Server Administrator's Guide.

For attributes related to migration, see the Cluster Server Bundled Agents Reference Guide.

Prev	Up	Next
Appendix A. Troubleshooting	Home	Live migration storage connectivity in a Red Hat Enterprise Virtualization (RHEV) environment