Here is my problem :
fleetctl list-units
UNIT MACHINE
processing-node@1.service X.Y.Z.86
processing-node@10.service X.Y.Z.150
processing-node@11.service X.Y.Z.48
processing-node@12.service X.Y.Z.48
processing-node@13.service X.Y.Z.48
processing-node@14.service X.Y.Z.86
processing-node@15.service X.Y.Z.82
processing-node@16.service X.Y.Z.48
processing-node@2.service X.Y.Z.248
processing-node@3.service X.Y.Z.48
processing-node@4.service X.Y.Z.85
processing-node@5.service X.Y.Z.48
processing-node@6.service X.Y.Z.48
processing-node@7.service X.Y.Z.48
processing-node@8.service X.Y.Z.87
processing-node@9.service X.Y.Z.248
worker-cache@1.service X.Y.Z.248
worker-cache@2.service X.Y.Z.222
worker-cache@3.service X.Y.Z.87
worker-cache@4.service X.Y.Z.150
worker-cache@5.service X.Y.Z.82
worker-cache@6.service X.Y.Z.85
worker-cache@7.service X.Y.Z.48
worker-cache@8.service X.Y.Z.86
The cluster is composed of ten machines. The worker-cache units needs a lot of calculation power so they exclude each other in the service file :
tail -2 worker-cache@.service
[X-Fleet]
Conflicts=worker-cache@*
So we have only one worker-cache unit per node. The processing-nodes units need less power and can be spawned on the same machines as the worker-cache units but I would like to have maximum two of them per machines which is definitely not the case actually :
processing-node@11.service X.Y.Z.48
processing-node@12.service X.Y.Z.48
processing-node@13.service X.Y.Z.48
processing-node@16.service X.Y.Z.48
processing-node@3.service X.Y.Z.48
processing-node@5.service X.Y.Z.48
processing-node@6.service X.Y.Z.48
processing-node@7.service X.Y.Z.48
Is there a way to do that ?
