[Linux-cluster] What is the proper procedure to reboot a node in a cluster?

Digimer linux at alteeve.com
Thu Mar 10 20:02:28 UTC 2011

On 03/10/2011 02:58 PM, Shi Jin wrote:
> Thank you.
> I think what you did makes perfect sense but on the other hand shouldn't
> the reboot process stop the services in the right order in the first
> place? Maybe there is a timeout issue or they don't necessarily follow
> the right order?
> Do you let the boot system to start the services for you in the right
> order or you actually have to do it manually? 
> Thanks.
> Shi 

It should, assuming that the KXXfoo entries exist for the cluster
services in the right order, and that they all shut down properly. Of
course, I find this not always is the case. So for that reason, I like
to stop things manually, when I have the luxury.

As for starting, it depends. If it's a machine I have ready access to, I
generally like to start things manually. Mainly because of how heavily I
use DRBD and it's tendency to have issues on startup. Not frequent, but
frequent enough. If you do automatically start everything, pay close
attention to the start order and do plenty of testing.

E-Mail: digimer at alteeve.com
AN!Whitepapers: http://alteeve.com
Node Assassin:  http://nodeassassin.org

More information about the Linux-cluster mailing list