| 2015-08-19 10:43:01 |
Romil Gupta |
description |
With the introduction of OVSvApp solution, we need to support fault management use-case to avoid data traffic loss.
In the kernel panic scenario, we will put the ESX host into maintenance or shutdown mode. |
With the introduction of OVSvApp solution, which consists of
a service VM hosted on ESX hosts within a data center.
To address the Fault Tolerance and High Availablity of OVSvApp.
OVSvApp solution caters to the following failure scenarios:
OVSvApp agent crash:
- upstart mechanism tries to bring up the agent back.
Open vSwitch process crash:
- Child process of OVSvApp agent will monitor the OpenvSwitch
and tries to restart it when crashed.
- This will help us to re-populate the flows at the earliest
to resume the tenant VM data network connectivity.
OVSvApp VM crash:
- Agent monitoring mechanism will put ESX host in maintenance mode
or powers off as per "esx_maintenance_mode" param configuration.
By default, 'esx_maintenance_mode' is True, so we put the host to
maintenance mode.
- vCenter DRS will migrate tenant VMs to other ESX hosts within the
cluster and safeguards traffic from being blackholed.
- Agent monitoring can be turned on/off using the configuration
"enable_agent_monitor" flag configuration in neutron configuration.
By default, it is enabled. |
|
