Synth runner is on, but no deployment is being created

Hi CrunchDAO team,

I’m trying to deploy my first Synth model (innovative-m / synth-alpha). The submission was accepted, but the runner never creates a deployment entry.

Here’s what I’m seeing:

  • Model ID: 16705
  • Submission ID: 75893
  • Hardware: CPU
  • Desired status: Start
  • Deployments: 0

I left it running for more than two hours. I also switched the runner off, waited for the blockchain confirmation, turned it back on, and clicked Deploy again. The UI showed “Deployment initiated”, but there is still no deployment row, run, or metric. The tracker passes its local smoke test, so it does not look like the container is reaching the point where the code could fail.

The on-chain model account is active and in START:

https://solscan.io/account/DbmPH2gDvKUePJ2uDYdKEErHjaZDRzGFPfiLPgahXmYr

Could someone please check whether Model #16705 is stuck in the runner provisioning queue and requeue or resync it if needed?

Thanks,
Mohamed

Hello innovative-m,

We were performing an infrastructure upgrade but one of our services failed to restart properly, and we failed to notice it.

It has been fixed, and I have ensured that your model is being deployed.

We apologize for the issue.

Hi Enzo,

Thanks, the deployment is now running and the model setup completes successfully.

I did a deeper check because almost five hours later, model #16705 still has no 1h prediction data. The Metrics API returns an empty dataset and no ranking, while the global Synth leaderboard is updating normally.

I also revalidated the exact submitted tracker across every current asset, horizon, and step. It produces valid, serializable predictions in 0.39 seconds, and the runner logs show no model error.

Could you please check whether #16705 was properly registered and synced with the Synth coordinator after the infrastructure restart, and whether it is actually receiving tick/predict calls? It may need a coordinator resync rather than another deployment.

Thanks,
Mohamed

Hello innovative-m,

I apologize for the delay. There was another issue with Pyth, a price provider, that cut off its API today…

This time, I made sure your model was getting called. Before, I only made sure it was deployed.

It will take up to 48 hours for your model to appear on the leaderboard and start seeing charts in the metrics section.