petals-infra / health.petals.dev

🏥 Health monitor for a Petals swarm
https://health.petals.dev
33 stars 28 forks source link

Invalidate cache if server state changes from JOINING to ONLINE #6

Closed borzunov closed 1 year ago

borzunov commented 1 year ago

The absence of version & cache info in the first 10 min after becoming ONLINE is known to be confusing for our users.

Also, we probably need to move this if inside the lock: https://github.com/petals-infra/health.petals.ml/blob/main/p2p_utils.py#L14

borzunov commented 1 year ago

Solved in https://github.com/bigscience-workshop/petals/pull/355, #7, and https://github.com/petals-infra/health.petals.ml/commit/04851559b578a5bc3330f9f7d076e7ae4242535e.