Monitoring

The Monitoring page answers, top to bottom: is anything on fire, what is the traffic doing, is the hardware coping, which app is responsible, and who is asking for what from where. Below the dashboard sits the full list of issues across your projects, servers and domains.

Controls

The bar at the top of the dashboard:

Your choice of range, server and auto-refresh is remembered in this browser. If a server's edge doesn't answer, a notice says so ("The edge isn't answering on N of these servers. Live numbers resume when it does.").

How often each part refreshes on its own: live numbers every 5 seconds; traffic history every minute (Live, 1h), 2 minutes (24h) or 5 minutes (7d, 30d); server metrics every minute (5 minutes for 7d and 30d); the apps table every 2 minutes (5 for 7d and 30d); issues every 2 minutes.

Server health dials

Server health shows three dials for the selected server, or Across N servers:

Each dial reads Normal, High or Critical (or No data) and marks Alert at the server's alert threshold. It turns red at the threshold and amber from 20 points below it, or from half the threshold when that is higher (amber from 70% for a 90% threshold). For all servers, the strictest threshold among them is used. The values are sampled every 5 minutes ("Sampled every 5 minutes"), and with several servers the busiest one is named ("Busiest: web-1"). Change the thresholds per server; see Server monitoring and alerts.

Headline numbers

Six tiles, each with a small sparkline and the change against the previous period of the same length ("vs previous 24h"; in Live, "vs last hour"):

All traffic numbers come from the edge, so they count requests to every domain the selected servers route.

Incidents

The Incidents strip shows open issues by severity, a timeline of incidents inside the selected window, and the most urgent open items. "All clear. Nothing open on these servers." means there is nothing to do. View all issues jumps to the list below.

Charts

Traffic:

• Requests per second by status: stacked 2xx, 3xx, 4xx and 5xx (in Live, 2xx/3xx, 4xx, 5xx), with the peak rate.
• Bandwidth: in and out.
• Response time percentiles: p50, p95 and p99. If the edge hasn't recorded response-time buckets yet, the chart says "Percentiles appear once the edge records response-time buckets."

Resources (from the 5-minute server samples):

• CPU, Memory and Disk, each with a dashed line at its alert threshold.
• Load average (1m).
• Network: received and sent.

Hover a chart to read exact values. An empty chart says "No traffic recorded in this window yet." or "No samples yet. Server metrics are recorded every 5 minutes."

Apps table

Apps lists every project over the selected window ("Every project, over the selected window"):

Click a column header to sort (by Req/s, highest first, by default). Click a row to open that app's Monitoring tab. Long lists show Show all N / Show fewer; metrics load for the first 40 apps ("Metrics load for the first 40 apps").

Breakdown

At the bottom:

• Top paths: the most requested paths. Only apps that have Top Paths turned on in their Monitoring tab are counted (see App monitoring).
• Responses: a donut of responses by status class over the range.
• Busiest domains: the eight domains with the most requests, each with its 5xx share.
• World map: requests by visitor country. Countries are counted per day, so the map covers whole days.

The issues list

Below the dashboard is everything that needs attention, from checks Synq runs in the background. Three tabs:

• Overview: open issues, with a summary By severity and By area, and a fleet card for self-hosted servers. Filter by Outage, Action needed or Advisory, search with Search monitoring, and press Re-scan to run every check now.
• Health: the live state of every deployed container (Check now, and the switch between automatic and on-demand checking).
• History: resolved container and server incidents, kept for 30 days.

Issues are grouped by area: Synq (this installation itself), Servers, Projects and Domains. Examples of what shows up: Deploy blocked, Routing not synced, Domain not verified, Certificate problem, Container down, Crash looping, Failing health checks, Server unreachable, Server disk almost full, Server memory high, Server CPU high, Edge down and Update available. Many come with a Fix or View button. "Nothing needs attention" means every check passed.

The same events can reach you by email, chat or push; see Notifications.

Related