Re: [gentoo-user] Re: kde4 upgrading

[Stroller]

I've edited your message when quoting it in order to meet my agenda.

On 28 Oct 2009, at 00:28, James wrote:
> ....
> PS, if one of you really smart guys figures out mass/parallel
> upgrades, then I'd use that, even set up my own server
> to keep it efficient. I'm not smart enough (not enough time
> at current mental aptitude) to set all of that up, unless
> somebody else does the foundational work.....
>
> But I very much like the concept. Upgrade a master system.
> Test it. Then  push your own binaries/files to the other systems
> you manage.

There are already a number of ways of managing multiple machines. How  
do you think universities, corporations and public bodies with  
hundreds or thousands of desktops manage? I think I would be looking  
at something like having the machines PXE boot a single image or NFS  
mounting various directories, if I were in your situation. I've never  
actually done this, but I'm sure a little research would produce a  
less labour intensive solution.

> ...
> Interesting, but not what I'm looking for. I do not mind
> upgrading the systems one at a time. I just do 1 per day,
> while I do other work. What has me "hacked" is that every time
> I do an upgrade to kde4, it seems to be a different set
> of problems, even though the upgrades are a few days apart.
> Multiply across a dozen workstations, and it's a time sink.

It seems to me, from your description, that your dozen machines are at  
the limit of your ability to maintain this way. No one would ever  
consider upgrading sites with 100 machines one by one each day, and it  
would be crazy to try and run a beefy thin-client server just to serve  
one or two desktops.

So the network has grown from a couple of machines to a dozen, and  
you're still doing things the same way - the question is, will you be  
able to continue doing things the same way if you were to double the  
number of PCs by next year?

I think that alternative methods of approaching system administration  
are sure to bring their own problems and require an investment of time  
to implement, but I don't see how upgrading machines one by one is  
sustainable. Honestly, it would be driving me crazy to be in your  
position, and I think some other alternative might well show time and  
hassle saved once it's up and running.

Stroller.