Re: [Evolution] Moving settings between computers

[guenther <guenther rudersport de>]

> > > > $ evolution --force-shutdown
> > > > $ gconftools-2 --shutdown
> > >
> > > This is a valuable tip... but... what are you supposed to do if you have
> > > to copy 10,000's of users from one NFS server to another, and you don't
> > > have access to all of the NFS clients, and no reliable way of getting in
> > > touch with all of the users?
> >
> > Uhm, this question isn't Evolution nor GConf specific...
> >
> > You are in trouble, sure. But you are always in trouble, when doing
> > this. Moving a lot of data and switching machines cannot be done, while
> > users are still accessing and modifying the data on the source machine.
> >
> > You are trying to move $HOME directories for thousands of users? Then
> > you need to make sure, no user is modifying his data while moving.
> > Otherwise, he might end up with data loss. You are allowing the users to
> > sort mails using procmail? Oh, then your MTA is accessing data in $HOME
> > as well. Likewise for any other app...
> >
> > Yes, "downtime" is a nasty word... ;-)
>
> I'm aware that problems can come from moving a bunch of $HOME's with
> active users, but from this discussion I've inferred rightly or wrongly,
> that gconf+evolution have changed this from a "look out for race
> conditions" problem to a "problems every time" scenario.
>
> I am worrying about nothing?  Or has the situation truly gotten worse?

Well, I don't know for sure.

But from my experience, syncing to disk isn't necessary most of the
times. Which does not mean, it isn't necessary or wise to do so. I
believe, if nothing changed since the last sync, it isn't necessary --
but it doesn't do any harm, and you're definitely on the safe side.

AFAIK it is about the same, as it was before.

On the other hand, shutting down the GConf daemon almost never was
mentioned in the past -- although necessary. Maybe that's the bit that
makes it sound more scary now?


Note, that the situation on the target side did not change either. As
long as a running process already has the data in memory, altering the
data on disk is worthless.

It is and was always a race condition between memory and disk data.
There are not "problems every time". In fact, it really works reliably.


That said: Now, that more apps start using e-d-s, this actually has more
potential to get difficult. The clock applet uses e-d-s to display
current appointments. I honestly don't know how this interacts with
e-d-s and if (or when) it will respawn e-d-s.


Please note, that these instructions are for moving/copying data over to
another machine. The mentioned servers are not designed with *syncing*
in mind, AFAIK. As this is meant as a one time process, logging out
isn't too much trouble.

MultiSync [1] can be used for syncing multiple Evolutions, IIRC. It
supports Evolution 2 now. :)  And AFAIK there are other projects out
there aiming at syncing Evo, which I don't remember right now. (Folks,
feel free to mention any apps here.)


Does that sound less frightening? Well, I guess this is an open
discussion, now... Let the *-fest begin. ;-)

...guenther


[1] http://multisync.sf.net


-- 
char *t="\10pse\0r\0dtu\0  ghno\x4e\xc8\x79\xf4\xab\x51\x8a\x10\xf4\xf4\xc4";
main(){ char h,m=h=*t++,*x=t+2*h,c,i,l=*x,s=0; for (i=0;i<l;i++){ i%8? c<<=1:
(c=*++x); c&128 && (s+=h); if (!(h>>=1)||!t[s+h]){ putchar(t[s]);h=m;s=0; }}}