Rebooting when patching

[vmproteau]

I've read differing opinions on this (even from VMWare support) and wanted to know if anyone has a definitive answer.

Of the 11 patches released January 2nd for ESX 3.0.2, 9 of them require a Host reboot. Is it valid to use the switch to supress the reboot for all of these patches and reboot once after all have been installed?

Are there any exceptions to this?

[virtualdud3]

I've always used the "-n" switch, and just rebooted once after the script finishes installing the patches.

As far as I know, there aren't any exceptions. If somebody does[/i] know of any exceptions, please correct me.

###############

If you found this information useful, please consider awarding points for "Correct" or "Helpful". Thanks!!!

[IB_IT]

we had issues with some of our 3.0.x servers where post patch scripts continued to run even after a reboot and causing problems with other services starting. We opened a support case with VMWare and the tech recommended doing a second reboot to the servers if you do multi patch installs with a surpressed reboot. I will see if I can dig that info up and post it here...

[dominic7]

I can see no technical reason why there would be any benefit from a second reboot.

we had issues with some of our 3.0.x servers where post patch scripts continued to run even after a reboot and causing problems with other services starting. We opened a support case with VMWare and the tech recommended doing a second reboot to the servers if you do multi patch installs with a surpressed reboot. I will see if I can dig that info up and post it here...

[IB_IT · Accepted solution]

Well, not sure what to tell you, dominic. Here is what the VMWare tech said about our issue:

"Some patches run post-reboot scripts, which change the behaviour of the daemons that are already running in the Service Console. Chaining multiple patches in one run might then require a second reboot. This is a known issue.

In your case a restart of the management services did the trick. To avoid incurring into troubles in the future I have some tips for you:

- the VirtualCenter Server gathers information from the a daemon running in each ESX host (called vpxa) which is an interface to the management service (the hostd deamon) which is an interface to the vmkernel running on the ESX host. These interfaces add 2 layers. To optimize the performance, data caching occurs. The caching can lead to inconsistencies because the data are not updated in real-time but only at certain intervals. If you are not able to perform a certain operation, it's generally advisable to restart the VirtualCenter Server and, if this does not clear up the problem, the management services in the affected ESX host.

- beware if you start/restart the management services, the virtual machines with automatic startup/shutdown settings will also be restarted.

- to stay on the safe side, when applying multiple patches which require rebooting the machine, I'd suggest to just reboot the ESX host a second time."

[vmproteau]

Most techs have told me it is OK to reboot at the end. This is the first I've heard about a 2nd reboot recommendation but, I may open a case and see if I can get more information. Regarding patches it's difficult to get consensus.

The only exception I've heard is for kernel patches. I had one tech who recommended rebooting after any kernel related patches before applying subsequent patches I suppose if you follow this recommendation you could just do the kernel patch last.