No relevant resource is found in the selected language.

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>

Reminder

To have a better experience, please upgrade your IE browser.

upgrade

Whether the HA virtual machine can execute the HA normally in the process of the sundry servers’ powering off?

Publication Date:  2012-10-23 Views:  39 Downloads:  0
Issue Description
Q: Cloud computing R1C01, R2C00 version. The virtual machine with the HA function can startup the HA function when the CNA node server is unmoral. Whether we can execute HA normally for all conditions of the sundry servers's powering off?
Alarm Information
None.
Handling Process
Test models of the following node servers’ powering off:
1. Run the operation on the OMS portal:
a) Click the power off button on the OmsPortal webpage, power off the server----all the virtual machine can execute HA normally.
b) Click the restart button on the OmsPortal webpage, restart the server----all the virtual machine can execute HA normally.
2. Run the operation on the BMC:
a) Execute the power off operation on the BMC webpage, power off the server----not all the virtual machine can execute HA normally.
b) Execute the restart operation on the BMC webpage, restart the server---- all the virtual machine can execute HA normally.
c) Execute the secure restart operation on the BMC webpage, restart the server----not all the virtual machine can execute HA normally.
3. Login in the CNA via putty to execute the command operation:
a) Login in the CNA via putty, execute the reboot command, and restart the server----not all the virtual machine can execute HA normally.
b) Login in the CNA via putty, execute the halt command, and power off the server----not all the virtual machine can execute HA normally.
c) Login in the CNA via putty, execute the halt-p command, and power off the server----not all the virtual machine can execute HA normally.
d) Login in the CNA via putty, execute the power off command, and power off the server----not all the virtual machine can execute HA normally.
e) Login in the CNA via putty, execute the init 6 command, and restart the server----not all the virtual machine can execute HA normally.
f) Login in the CNA via putty, execute the init 0 command, and restart the server----not all the virtual machine can execute HA normally.
4. Run the operations directly for the server hardware:
a) Cut off the power supply of the CAN node directly----------------------------------------all the virtual machine can execute HA normally.
b) Press the server’s power off button to shut down----------------------------------------- all the virtual machine can execute HA normally.
5. Use the watchdog to restart the server:
a) Login in the node via putty, cause the key process fault, use the watchdog to restart the server-------- all the virtual machine can execute HA normally.

Conclude from the testing exercise, we find only the following six conditions in which the virtual machine can execute HA successfully, and the other powering off, restarting are can’t ensure all the virtual machine execute HA successfully:
OmsPortal power off, OmsPortal restart, BMC restart, cut off the power supply, press the server’s power off button, watchdog restart.
Root Cause
None.
Suggestions
1. If you need all the virtual machine can startup the HA when the server is powering off, you must stop the processes on the CAN in turn, the NC process must be stopped after the watchdog process has been stopped (as in the OmsPortal power off condition, OmsPortal restart condition and watchdog restart condition), or the server cut off the power supply instantly and all processes are shut down (as in the BMC restart condition, watchdog restart condition and press the server’s power off button condition).
2. Run the power off operation in some other levels of Operating System, the system stops all the processes firstly, but it can’t insure that the watchdog process is stopped before the NC process has been stopped, this may cause the following result:
a) The system stops the NC process firstly, then stops the virtual machine, but the watchdog has not been stopped.
b) The NC process will be started by the watchdog process, and then the NC process will report to the CRM with the message that the Stopped virtual machine is in the STOPPED state, this will cause these virtual machines can’t execute the HA.
c) Before the CNA server run the operation to close the system, the interval of NC process refreshes the state is very short, this may cause not all the stopped virtual machines ‘STOPPED state can be reported to the CRM, these un-reported virtual machines can execute the HA normally.
3. Consequently, when you want to run the power on/off operation for the server, use the operation command button on the OMS Portal as far as possible; don’t login in the server operating system to execute the operating system command for the server.

END