srl-labs
Host resource verification checks per node not per lab
Some kinds (srl, sros, cjunosevolved) have a way of verifying the host has enough resources for the node. This check however seems to only check on a per node and not that the host has enough resources to deploy the whole lab.
This, for example, that if my system has 2gb of ram and 2cpu cores available containerlab will allow me to deploy a 100SRL node lab.
yes, this is a little bit intentional. Resources are set per-node as this is how most vendors (if at all) put them on paper. This, however, a recommendation and people manage to squeeze in way more on way less, which is how most labs are run - oversubscribed.
While we potentially could calculate the sum of recommended resources when users start up a lab and issue a warning (we won't prevent the start), this would mostly be inaccurate approximation.
Is it a practical challenge to apply common sense and not schedule 100 NOSes on raspberry pi?
