[Hcoop-discuss] Concerns regarding the new server plan

[Adam Chlipala]

rob at hcoop.net wrote:

>It feels like it was just a month ago that we switched from Abu to Fyodor.
> It's unfortunate we weren't able to anticipate this need for a better,
>more capable, server prior to our migration from Abu.
>  
>
I thought my plans about this had been clear all along.  I was planning 
to move again as soon as possible the whole time, and I thought that the 
general membership consensus hovered around this idea, too.

Our Xiolink-based hosting set-up had many flaws that made it 
outrageously untenable, especially in the face of increasing 
membership.  I've had a consistent idea of what our ideal hosting set-up 
looks like; the issue is building up a large enough membership base to 
make its costs reasonable.  We couldn't support enough members on 
Abulafia to reach that point, but I think we're in a position to reach 
it in the near future, if we're not there already.

>I am concerned that because of the potential growth of HCoop this process
>of integrating new servers and migrating user data to these servers could
>happen rather frequently.  Or, at least frequently enough that it feels
>inconvenient for me to need to go through steps similar to what we did
>last time (http://wiki.hcoop.net/wiki/MemberMigration).
>  
>
I hope that our next set-up will be our primary home for several years, 
at least.  One of its main features, in my mind, is a central shared 
filesystem, which should make adding new servers and decommissioning old 
ones a relative breeze.  There may still be some annoying manual work to 
transition to it from our current set-up, since filesystem structure 
will probably be reshuffled.

>Is anyone thinking of a way to migrate user data that will have less
>impact on the members?
>
I have a suspicion that by "anyone" you mean "any of the volunteer 
admins," and that you expect us to implement as well as come up with 
ideas.  Keep in mind that we're just that, volunteers, and the fact that 
we do what we do indicates that we have relatively high pain thresholds 
as far as rare administrative intervention effort goes.  If this is an 
important issue for you, then you will probably need to be involved in 
planning and implementation if you want to be sure that your desires are 
met... and that will probably be more work than you would have to do for 
this final migration!