[Linux-HA] Groups vs colocations.... etc

[Andrew Beekhof]

On 12/12/06, Andreas Kurz <akurz at sms.at> wrote:
> Andrew Beekhof wrote:
> ...
> > because they're just a hardcoded shortcut and although they work fine
> > in many cases, as soon people want to do anything a little more
> > tricky/interesting then groups dont help anymore (and actually make
> > things worse because its not obvious that they wont work)
>
> So groups are only a shortcut and they make the config more readable.

correct

> You also mentioned in some posts everything what a group does can be
> made by some constraints by hand. But then I don't understand what's the
> difference.

because they are less flexible

> What do groups more in the background than making constraints?

there are a huge number of ways a series of resources can potentially
be combined and related to each other.  for them to be equivalent, I
would need to anticipate every possible arrangement of resources and
their options.  naturally thats not feasible and so groups can only
ever be a subset of what is possible "manually".

>
> ...
> >> Yes, that is clear but like groups make all co-location constrainst
> >> automatically,
> >> I thought it would be a convient way to pass my node specific
> >> attributes to all
> >> resources in the group.
> >
> > and it will for everything except the two stickiness values because
> > they're calculated differently to allow even this much to work
>
> OK
>
> ....
> >
> > if you update again you should find it working a little better
>
> A little bit ;-) ... now the scores are added again for all child
> resources, but only for the node the resources are currently running on.
>
> Should a resource with a negative score not result in the whole group to
> be migrated away if there is another node available?

in a word no.  because the group as a whole may still prefers to run there.
can you update again, i fixed some more group-related functionality
late last night.