# App monitoring
URL: https://synquic.tech/docs/guides/app-monitoring.md
A project's Monitoring tab, its live resource use, usage history, traffic panel with latency percentiles, visitor map, top paths and responses, plus the Health section that records container failures.
Each project has a **Monitoring** tab that shows how that one app is doing: what it uses right now, how that
changed over time, how much traffic it gets and from where, and what went wrong with its containers.
Open the project and choose **Monitoring** in the project menu (`/projects//monitoring`). On a self-hosted
project it has two sections, **Monitoring** and **Health**. The same data feeds the **Apps** table on the
[Monitoring page](/docs/dashboard/monitoring#apps-table); clicking an app there opens this tab.
- A deployed project. A brand-new app with no traffic shows **No monitoring data yet** until the first requests
arrive.
- For traffic numbers, a domain served by the Synq edge. Request counts come from the edge, not from your code.
## Headline numbers
A strip of four numbers for the selected period:
| Number | Meaning |
| --- | --- |
| **Server Requests** | Requests the edge served for this project. |
| **Visitor-days** | Distinct visitors, counted once per day. The total is an upper bound, since a returning visitor counts once per day; hover it for the busiest single day. |
| **Avg Response** | Mean response time. |
| **Bandwidth Out** | Data sent to visitors, with data received underneath. |
## Domain scope
When the project has more than one domain, a picker scopes the whole tab (**All domains** by default, or one
domain). The traffic panel, map, top paths and responses all follow it.
## Resource usage
**Resource usage** ("Live, refreshed every few seconds") shows **CPU** (of the cores it may use), **Memory**
(of its limit), **Network** and **Disk I/O**, with a **Live** / **Disconnected** marker and **Reconnect** if the
stream drops. For a project with several services, buttons in the header switch between **All services** and
one service; the choice also scopes **Usage over time** further down.
## Traffic panel
On a self-hosted project the detailed traffic panel follows ("Requests the edge served for this project"). A
project on Synq Cloud shows a collapsed **Traffic overview** chart instead.
**Live**, the last 60 seconds, refreshed every 5 seconds:
| Tile | Meaning |
| --- | --- |
| **Active requests** | Requests in flight right now. |
| **Requests/s** | Rate over the last 60 seconds. |
| **Inbound** / **Outbound** | Bandwidth per second. |
| **Error rate** | Share of `5xx` responses: amber from 1%, red from 5%. |
| **Avg response** | Mean response time over the last 60 seconds. |
**Over a range**: choose **1h**, **24h** (default), **7d** or **30d**. A summary row shows **Requests**, **Avg
rate**, **Peak rate**, **Data in**, **Data out**, **5xx rate** and **4xx rate**, followed by three charts:
- **Requests per second**, split by status class;
- **Bandwidth**, in and out;
- **Response time**: p50, p95 and p99. "Percentiles appear once the edge records response-time buckets."
means the edge hasn't collected latency data for this window yet.
Use it to answer questions like "did the error rate go up after the last deploy?" (compare the `5xx` band in
the requests chart with the deploy time) or "is the app slow for everyone or only some requests?" (a high p99
with a normal p50 means only some requests are slow).
## Visitors by country
The world map shows requests by visitor country ("N requests from M countries"), with a donut and legend
underneath. Countries are counted per day. Controls on the map:
- **Theme** / **Flags**: colour the countries by theme colour or by flag.
- **Live**: animate each incoming request on the map, from the same stream the **Logs** tab shows.
- **Recent hits**: show the list of recent requests next to the map.
If the map stays empty with "Country data unavailable on this edge", the edge can't resolve visitor countries;
traffic totals are unaffected.
## Top paths
**Top Paths** lists the most requested paths. It is the one metric that is off by default, because counting
per path is about 57% of the work the edge does on every request; everything else on this tab is collected at no
extra cost. Press **Enable Top Paths** to turn it on for this project, and **Turn off Top Paths** to stop.
Only projects with Top Paths on contribute to the **Top paths** list on the Monitoring page.
## Responses
**Responses** breaks the requests down by status code (for example `200`, `304`, `404`, `500`), with rare codes
grouped under **Others**.
## Usage over time
At the bottom, **Usage over time** charts CPU and memory from samples taken every 5 minutes. The first points
appear a few minutes after the app starts running, and samples are kept for 30 days.
**Volume disk usage** ("Measured hourly") charts the size of the project's volumes, when it has any.
## Health
The **Health** section ("Container failures recorded for this project by the health watch") lists:
- **Open incidents**: problems happening now, with how long they have lasted ("ongoing for 12m");
- **Recent**: incidents of the last days, with how long each lasted.
Each incident has a kind (**Down**, **Crash loop**, **Unhealthy** or **Server unreachable**) and, when known,
the exit code, the number of restarts, **out of memory** for a container killed for using too much memory, and
**Show log excerpt** with the container's last log lines.
The records come from the **Container health watch** job, which checks every server's Docker containers once a
minute. If it's turned off, the section warns "The container health watch is off" with **Turn it on in Jobs**:
an empty list then doesn't mean everything is healthy. The same incidents trigger the **App down**, **App crash
looping**, **App unhealthy** and **App recovered** notifications (see [Notifications](/docs/guides/notifications)).
## Related