Fly.io

Ship logs and platform metrics from your Fly.io apps to Parseable. Both pipelines are dedicated Fly apps deployed alongside (not inside) your workload apps; your existing fly.toml files are not modified.

Overview

Pipeline	Source	Shipper	Parseable route	Availability
Logs	Fly's NATS log stream (`nats://[fdaa::3]:4223`)	`fly-log-shipper` with Parseable Vector sink	`POST /api/v1/logstream/...`	All Parseable editions
Metrics	Fly's managed Prometheus (`api.fly.io/prometheus/<org>/federate`)	`parseable-fly-metrics` (Go agent)	`POST /v1/prometheus/write`	Parseable Enterprise / Prism only

One agent per Fly org covers every app in that org - no per-app config.

Logs

Architecture

your-app-1 ─┐
your-app-2 ─┤── Fly NATS logs stream ──▶ fly-log-shipper ──▶ Parseable
your-app-N ─┘    (nats://[fdaa::3]:4223)    (Fly app)         /api/v1/logstream/<stream>

Deploy

fly-log-shipper is Fly's official log forwarder, with a Parseable sink available in the upstream repo (see PR #76).

git clone https://github.com/superfly/fly-log-shipper
cd fly-log-shipper

# create the shipper app in your Fly org
flyctl apps create <org>-fly-log-shipper --org <your-fly-org>

# auth + Parseable sink secrets
flyctl secrets set \
  ACCESS_TOKEN=$(flyctl auth token) \
  ORG=<your-fly-org-slug> \
  PARSEABLE_URL=https://<your-parseable-host> \
  PARSEABLE_USERNAME=<user> \
  PARSEABLE_PASSWORD=<pass> \
  PARSEABLE_STREAM=fly_logs

flyctl deploy --ha=false

Logs from every app in the org now flow into the fly_logs stream. Refer to the fly-log-shipper README for the full list of sink env vars.

Metrics

Pro Enterprise

The Fly metrics pipeline writes via Parseable's POST /v1/prometheus/write route, which is available in Parseable Enterprise and Prism (distributed) deployments. The open-source Parseable server does not expose this route.

Architecture

your-app-1 ─┐
your-app-2 ─┤── Fly managed Prometheus ──▶ parseable-fly-metrics ──▶ Parseable
your-app-N ─┘    api.fly.io/prometheus/<org>/federate    (Fly app)     /v1/prometheus/write
                                                                       (ingestor node)

Deploy

git clone https://github.com/parseablehq/parseable-fly-metrics
cd parseable-fly-metrics

# edit fly.toml - set `app = "<your-unique-name>"` and `primary_region`

flyctl apps create <your-unique-name> --org <your-fly-org>

flyctl secrets set \
  FLY_ORG=<your-fly-org-slug> \
  FLY_METRICS_TOKEN=$(flyctl auth token) \
  PARSEABLE_URL=https://<your-parseable-ingestor>:8010 \
  PARSEABLE_USERNAME=<user> \
  PARSEABLE_PASSWORD=<pass>

flyctl deploy --ha=false

Expected log line per scrape interval:

scrape ok: families=61 series=134 samples=134 duration=460ms

PARSEABLE_URL must point at a Prism ingestor node, not the query node. Run a single agent machine per org; multiple instances would double-scrape.

Configuration

Variable	Required	Default	Notes
`FLY_ORG`	yes	-	Fly org slug. Find via `flyctl orgs list`.
`FLY_METRICS_TOKEN`	yes	-	Output of `flyctl auth token` or `flyctl tokens create readonly -o <org>`.
`PARSEABLE_URL`	yes	-	Base URL of the Parseable ingestor.
`PARSEABLE_USERNAME`	yes	-	HTTP basic auth username.
`PARSEABLE_PASSWORD`	yes	-	HTTP basic auth password.
`PARSEABLE_STREAM`	no	`fly_metrics`	Target stream name. Parseable creates the stream automatically on first write.
`SCRAPE_INTERVAL_SECS`	no	`30`	Seconds between scrapes.

Verify

Query the Parseable query node:

curl -u <user>:<pass> -X POST \
  -H 'Content-Type: application/json' \
  -d '{"query":"select metric_name, app, region, data_point_value from fly_metrics order by p_timestamp desc limit 5","startTime":"1h","endTime":"now"}' \
  https://<your-parseable-query-node>/api/v1/query

Sample row from the fly_metrics stream:

{
  "app": "hellofly",
  "region": "sin",
  "instance": "9080d161a52928",
  "metric_name": "fly_instance_memory_buffers",
  "metric_type": "gauge",
  "data_point_value": 778240,
  "time_unix_nano": "2026-05-18T06:45:38.637",
  "scope_name": "prometheus_remote_write",
  "p_format": "otel-metrics"
}

For logs, query the fly_logs stream the same way.

Notes

Auth scheme. Fly's Prometheus endpoint requires Authorization: FlyV1 <token> for tokens issued by recent flyctl versions. The agent handles this automatically; the Bearer scheme returns HTTP 401.
Idle apps. Apps with auto_stop_machines = "stop" stop emitting metrics until they are woken. Fly's /federate returns HTTP 200 with an empty body in that state.
Self-observability blind spot. Both agents (logs and metrics) are themselves Fly apps in the same org, so they appear in their own data. If an agent crashes, its own outage is invisible from Parseable; set a stream-silence alert as a watchdog.

Fly.io

On this page