[Pacemaker] Question about symmetricity of the colocation constraints

[Andrew Beekhof]

On 7 May 2014, at 10:03 am, Jacob Nikom <JNikom at kivasystems.com> wrote:

> Hi,
> 
> I have question about colocation constraints – are they symmetrical (commutative) or not?
> 
> What I mean is the following.
> 
> If we have the colocation constraint like
>   <rsc_colocation id="resource1-with-resource2" rsc="resource1" with-rsc="reource2" score="-INFINITY"/>
> 
> Does it produces the same behavior as the constraint
>   <rsc_colocation id="resource2-with-resource1" rsc="resource2" with-rsc="reource1" score="-INFINITY"/>

Almost.  
They wont be colocated in either case, but (using the second case as an example) in order to know where resource2 cant go, we need to know where resource1 is (or is going to be).
So resource1 will get its highest preferred node and in a situation where both cannot be active, resource1 will win.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20140507/ce678cad/attachment-0003.sig>