How Intelligence dashboards work
Each integration sends specific metrics to AppSignal. When AppSignal receives the trigger metric of a known integration, it creates a dashboard with charts tailored to that integration. Dashboards are created per app environment, so a staging and a production environment each get their own. An Intelligence dashboard carries an Intelligence badge in its header. In the sidebar, a wand icon marks a type of Intelligence dashboard you have not opened yet. Dashboards that are still being refined carry a beta badge as well; their charts and metrics may change. The first time AppSignal creates a type of dashboard for you, it emails the members of the app. You can turn these emails off in your notification settings. Dashboards are designed to answer common operational questions quickly. Health and error indicators appear first, followed by resource utilization, then operational context. Chart descriptions explain what values indicate a problem and what to do about it.Customizing an Intelligence dashboard
An Intelligence dashboard works like any other dashboard. You can add charts, edit the existing ones, rename it, and delete it. Renaming does not count as a customization. Deleting an Intelligence dashboard is permanent for that app environment. AppSignal does not create it again when it sees the metrics later.Dashboard updates
AppSignal improves Intelligence dashboards over time: charts are added, metrics change, and descriptions are refined. When a newer version of a dashboard is available, the dashboard shows the message A dashboard update is available with three options:- Update this dashboard replaces the charts with the latest version. Any customizations you made are overwritten; the dashboard’s name is kept.
- Create a new dashboard instead creates a fresh Intelligence dashboard next to your current one. Your current dashboard keeps its charts, gets a “(custom)” suffix, and no longer receives updates.
- Skip for now hides the message until the next version is available.
Available Intelligence dashboards
Intelligence dashboards are available for the following integrations, grouped by category. Each link leads to the integration’s setup instructions and dashboard details.AWS CloudWatch
AppSignal supports 46 Intelligence dashboards for AWS services through CloudWatch Metric Streams. Each AWS service gets its own dashboard with charts ordered by urgency of signal. Services with more than one flavor, such as EKS, ElastiCache, and Amazon MQ, get one dashboard per flavor. An AWS/CloudWatch dashboard tracks the health of the metric stream itself: metric updates sent and the publish error rate. These dashboards currently carry the beta badge. See AWS Intelligence dashboards for the full list of supported services and their charts.Ruby
The following integrations create Intelligence dashboards when AppSignal detects their metrics in a Ruby app:- Active Job — job duration, job status per queue, and throughput per job class.
- Sidekiq — worker duration, queue latency, queue length, job status, Redis memory usage, and connection counts.
- Puma — thread pool capacity, connection backlog, and worker status.
- MongoDB — query throughput and query duration per database.
- Ruby VM — garbage collection counts and time, heap slots, allocated objects, Ruby VM statistics, and thread count.
- Global VM Lock — GVL wait time and waiting thread count.
- Action Mailer — deliveries per mailer and action, as a total and as a percentage of all deliveries.
- Karafka — consumed messages and batches, lag, errors and dead letter queue, connection stability, and process and thread counts.
Elixir
- Oban — job duration, queue wait time, and job counts broken down by worker, queue, and status.
- Ecto — query time, throughput, total time, decode time, queue time, and idle time per repo.
- Erlang VM — I/O, schedulers, processes, memory, atoms, scheduler utilization, and run queue length.
Node.js
- Heap statistics — V8 heap size, used heap, malloced memory, and native and detached context counts.
General
- Process Memory — average and 95th percentile RSS memory usage per host and process type. Available for Ruby and Elixir apps.
Infrastructure
- NGINX — request time, throughput, status codes, connections, and upstream proxy metrics.
- Kubernetes — two dashboards: node CPU, memory, swap, disk, inodes, and network; pod CPU, memory, and swap.
Heroku
Heroku creates several Intelligence dashboards from Log Drain or Telemetry Drain data:- Status codes — HTTP status code breakdown (Cedar generation only).
- Redis — connections, hit rate, evictions, memory, load average, and I/O.
- PostgreSQL — database size, connections, cache hit rates, memory, load average, I/O, WAL usage, and follower lag (Standard and Premium plans).
Render
Render creates Intelligence dashboards from a Metrics Stream:- HTTP — request volume and latency per service.
- Postgres — connections, storage size, transactions, table scans, locks, and replication lag.
- Key Value — connections against the connection limit.
Vector
AppSignal creates dashboards from metrics collected through Vector sources:- PostgreSQL — row operations, transactions, deadlocks, connected backends, and temporary file writes.
- MongoDB — three dashboards covering status (memory, connections, operations), WiredTiger storage engine internals, and replication health.
Vercel
- Speed Insights — the Web Vitals dashboard: TTFB, FCP, LCP, FID, and CLS from real-user measurements.