Monitoring
Know it's broken before your users do.
Every agent reports CPU, memory, disk, network and container health out of the box. Add uptime checks for your endpoints and send alerts to the people who can fix them.
A chart of disk usage on nyc-db-01 over two hours with an alert rule at 90 percent. A four-minute spike above 90 percent is ignored. Later, usage stays above 90 percent, and after ten minutes an alert is sent to the ops channel.
Alerting
From a metric to the right person.
Rules decide what's worth waking someone for. Routes decide who that is.
-
Agent
samples every 10s
CPU, memory, disk, network and container health from every server.
-
Rule
disk.used > 90% for 10m
A threshold and a duration, so a short spike doesn't page anyone.
-
Route
owner: ines · quiet 22–07
Who hears about it, and whether it can wait until morning.
-
Channel
slack #ops · webhook
Slack, email, webhooks, or your on-call tool on Enterprise.
Uptime
Checks from three continents.
HTTP, TCP and certificate checks run from Europe, North America and Asia. An endpoint is only marked down when more than one region agrees.
A monitoring view with 45 days of uptime bars for three endpoints, a CPU chart for the fleet compared with last week, and an alert about high disk usage on nyc-db-01.
Out of the box
Collected without installing anything else.
Servers
- CPU and load average
- Memory and swap
- Disk usage and I/O
- Network in and out
- Uptime and reboots
- Agent connection
Containers
- CPU and memory per container
- Restarts
- Out-of-memory kills
- Health check results
- Image and release
- Exit codes
Endpoints
- HTTP status and latency
- TCP reachability
- TLS certificate expiry
- Checks from 3 regions
- Response body match
- Every 30s to 10 min
Questions about monitoring
Do I still need Prometheus and Grafana?
Not for server and uptime basics. If you already run them, keep them: Fwoop's webhooks and API work alongside.
Can I monitor servers that don't run the agent?
Uptime checks work for any public endpoint. Server and container metrics need the agent.
Where are metrics stored?
In the control plane, for your plan's retention period: 7, 30 or 90 days. Logs and application data stay on your servers.
Can I silence alerts during maintenance?
Yes. Schedule a maintenance window for a server, a label or the whole fleet, and hold non-urgent alerts until working hours.
Bring your first server online in five minutes.
Start free with three servers. No credit card, no sales call, and your workloads stay on machines you own.