Topic: Good advice on how to handle the required reboot after upgrades?

[holck]

I have been running an SME server for a long time and also frequently installed updates and new software. As many of you will know, after an upgrade you often receive this message:

Code: [Select]

WARNING: You now need to run BOTH of the following commands
to ensure consistent system state:

signal-event post-upgrade; signal-event reboot

You should run these commands unless you are certain that
yum made no changes to your system.
But rebooting the server is a bit problematic for me. I have something like 20 users, and must of course announce a reboot first and find a convenient time to do it.

Do any of you have good advice on how to handle these situations? Do you just keep the server running and hope for the best? Or do you postpone updates until a time (maybe during a weekend), where you can go through the upgrade and reboot without too much hassle? Or do you (like I often do) try to guess which templates you should expand, and which services to restart?

Kind regards,
Jesper, Denmark

[brianr]

Do any of you have good advice on how to handle these situations? Do you just keep the server running and hope for the best? Or do you postpone updates until a time (maybe during a weekend), where you can go through the upgrade and reboot without too much hassle? Or do you (like I often do) try to guess which templates you should expand, and which services to restart?

That is what I do - late at night or over a weekend when (you hope!) no-one will be using the system.  I access remotely of course.

[guest22]

I agree with Brian, but would take it a step further by designing a maintenance policy and make sure your users are aware of this policy. Not just reboots should be incorporated but all service effecting events such as internet connections, routers, printers etc etc..

[Jean-Philippe Pialasse]

it is not what dev would recommand but usually when I do an update i simply expand the correct templates and restart the afflicted services.

You must also pay attention if there is not migrating informations that should be executed before that in  /etc/e-smith/db subfolders

Then you can update the config db to remove the anoying server-manager message.

If you have no modifications like compiled module, kmod etc, you can also try the update at night for the reboot, and cross your finger everything will reboot fine, as you will have to wake up early to check what goes wrong otherwise.

I usually reboot when a new kernel is available or a lot of services have been updated at once.

Remember rebbot will stay the safer way.

Reboot after update is a current problem for admin, it has been a choice done by the dev team as a lack of ressources to handle all the cases to avoid this. The dev team is still missing ressources to work on this, so if you have spare time or money


Update are not necessary that often, so a policy with warning 1 day prior to the update should ba also a good idea.

[Brave Dave]

I agree with Brian

With Samba especially, running the Yum update is going to risk putting that process in a state of disarray

If the process is a database type - MyOB is one that comes to mind, Quicken another - then you risk corrupting that database

I my case, mostly my clients use MyOB to pay me

Therefore running Yum could directly affect my beer on Friday night (not Joking !)