What happens in production. Live.
Container logs, CPU and memory every 2 seconds, email alerts, autoscaling and a terminal inside the container, all from the service panel.
At a glance
- Logs
- Live, 500 lines on opening
- Metrics
- CPU and memory, every 2 s
- Alerts
- By email, on a threshold
- Autoscaling
- CPU-based, adjustable min and max
See, get warned, act.
01
Live logs.
The Logs tab streams the container's output in real time: the last 500 lines on opening, then the stream goes on.
Scroll up to read calmly, scroll back down to follow new lines again.
02
CPU and memory, every 2 seconds.
Two charts, relative to the limit allocated to the service: you see right away if it is short on headroom.
Every build's logs sit next to them, to compare a working version with the one that breaks.
03
Email alerts.
One rule per metric (CPU, memory, disk or network): a threshold, a duration, and a delay before the next alert.
When the threshold is crossed, the email goes out. Rules can also be managed through the API, with a key limited to alerts.
04
Autoscaling.
Set a minimum, a maximum and a CPU target, 70% for example. Above it, an instance is added; when the load drops, it goes away.
The maximum stays capped by your plan. With several instances, keep sessions in Redis or the database.
05
A terminal inside the container.
A shell opened in the running container, with its files and variables, no SSH and no setup.
It is for diagnosing or running a migration: a redeploy starts from the image, so changes made by hand do not stay.
06
The platform's status, in public.
The /status page shows the state of every component, availability history, incidents and announced maintenance.
An incident shown there is already being handled by the team: its thread is updated until it is resolved.
What it does not do yet.
Autoscaling follows CPU only, not memory or request count.
Alerts go out by email; no webhook or chat integration yet.
An open terminal closes when the service is redeployed or restarted.
A volume is attached to one instance: with autoscaling, keep shared files elsewhere.
Questions about monitoring
The rest of the platform
Watch your service run.
Logs, metrics and alerts are ready from the first deployment.