Odysseus Odysseus

The platform now handles scheduling and dispatch, so you no longer need a container running just to wait for the right time. Pick a destination from your project’s services, write a payload using placeholders like {{ run_at }}, {{ worker_id }}, and {{ run_id }}, and we deliver on time with automatic retries, timeout control, and a full run history. Watch every run, trigger one manually whenever you need to, and see your cron workers right alongside your other services on the project map. Every run carries trace context, so a failure becomes a needle that’s easy to find.

When a request flows through your backend, hits your Postgres, warms your Redis, and fires an event on NATS, the trace now shows each leg of the trip as its own lane, with a readable name and per-service timings. Calls to the internet and to external destinations (Stripe, OpenAI, S3, any host outside your project) show up as their own nodes on the trace’s mini service map, one for the internet and one per external host. Click a lane to filter the waterfall to that service, or jump straight to the destination service’s detail page. The trace list also gained a chip showing how many services participated in each request.

Every service now has a Profiling tab with an interactive flamegraph, no agent to install, no dependency to import, no rebuild. Automatic support for Node.js, Java, Go, Ruby, and Rust. There’s a dedicated Profiling Explorer under Operations for longer investigations, and every trace span gets a Profiling tab filtered exactly to that span’s time window, perfect for going from “this request was slow” to “this loop was burning 40% of CPU” in a few clicks. Retention respects your plan, and the flamegraph lives inside the same visual identity as the rest of the dashboard.

Bring up services from images in private external registries without an intermediate pipeline. Drop the registry URL, username, and token into the service wizard (or the Settings tab later) and we handle authentication on every pull. A Test connection button validates your credentials before save, with clear messages when something’s off, expired certificate, wrong host, token without read access. If you’d rather, save without verifying, the action is recorded in your audit trail. Public images on build_method='image' keep working without any credentials.

Your audit trail now shows the outcome of each action (success or failure) and lets you filter by it, perfect for hunting failed logins, configuration changes that didn’t go through, or actions blocked by permissions. In parallel, we hardened the layer that records these events so tokens, passwords, and API keys are stripped before they ever land in a log, regardless of casing (camelCase, kebab-case, or snake_case). You see what happened, without ever seeing what you shouldn’t.