Skip to content

Conversation

@pkazmierczak
Copy link
Contributor

@pkazmierczak pkazmierczak commented Nov 21, 2025

This PR is an experiment in using TLA+ to model a high-level specification of the system scheduler. For now, we focus on modeling transitions in the state machine and provide a very basic temporal formula:

$$\text{SystemCoverage} \;\triangleq\; \Diamond\left( \forall j \in \text{currentJobs}:\; \forall c \in \text{eligibleNodes}:\; Eligible(j,c) \;\Rightarrow\; allocs[j][c] = 1 \right)$$

which means that for every job, every node will eventually have exactly one alloc. We provide some other basic safety properties like making sure nodes don't exceed capacity.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant