Observability,
out of the box.
Logs, metrics, traces and deploy events - wired into every service the moment it ships. Plus first-class bridges to Datadog, New Relic, Grafana and the stack you already pay for.
A deploy
without signal
is a guess.
You shipped. Did it work? Logs are in one tool, metrics in another, traces nowhere. The deploy marker never made it to the dashboard. Observe is the layer that closes the loop.
Instrumentation that ships itself
No Datadog agent DaemonSet to debug. No sidecar to forget. Qovery installs and maintains the collector on every cluster - OpenTelemetry out, nothing proprietary locked in.
Signals joined to deploys
Latency climbed at 14:02? Qovery already knows the 14:01 release is the cause. Every chart has the deploy markers. Every incident has a one-click revert.
Use what you already pay for
You have a Datadog contract. Or Grafana Cloud. Or Splunk. Qovery does not replace it - it fans out to it, preserving tags, costs and dashboards.
Signals,
without the YAML.
Six capabilities that turn a raw cluster into a service you can actually see.
Every container's stdout/stderr ingested, indexed, searchable. Structured JSON detected. Retention per environment class.
Node, pod, container, HTTP, DB - Prometheus-compatible. Kept for 30 days in-platform, forwarded to your tool for long-term.
OpenTelemetry auto-instrumentation for Node, Python, Go, JVM, .NET. Toggle on; traces appear.
Every release, config change and infra mutation is a timestamped event on every chart.
Define SLOs per service. Burn-rate alerts out of the box. Route to PagerDuty, Opsgenie, Slack, webhook.
Per service, per environment, per tenant. Owned by the team that owns the service.
Your tools,
still your tools.
Observe does not ask you to rip and replace. Forward any signal to the vendor you already have a contract with.
Logs, metrics, traces forwarded with Qovery tags and deploy events.
NRQL-compatible forwarding. Qovery services appear as entities automatically.
Prometheus remote-write, Loki logs, Tempo traces.
HEC ingestion. Log-only or full OTel pipeline - your choice.
Native wire format. Point any OTel endpoint at Qovery.
Scrape your existing Prometheus or let Qovery scrape your workloads.
Incident routing, auto-resolve on deploy fix, service-ownership mapping.
Deploy notifications, SLO burn alerts, incident summaries.
The agent
reads the graph.
A p95 spike, a cost cliff, a dependency timing out - the Agentic layer correlates signals with deploys, proposes a root cause, and offers the fix.
Detects regressions by itself
The agent watches every signal and flags the ones that matter.
Explains the graph
Ask why. The agent walks the dependency chain, cites the spans that moved.
Acts, safely
Suggested revert, suggested scale-up, suggested circuit breaker - ready to apply, one approval away.
"We had Datadog. We had Grafana. We had traces no one looked at. Qovery didn't replace any of it - it joined them to our deploys, so when latency moves we can tell why in seconds instead of hours."
Ship it.
Then see it.
Every service gets logs, metrics, traces and deploy markers on day one - forwarded to the tool your team already uses.