Logs and metrics

See what your app is doing without opening the AWS console.

Logs

The Logs tab of a deployment shows three kinds of lines:

Source	What it is
App	Everything your app writes to standard output and standard error. Use a console log handler.
Build	The output of building your image, including the [djangocloud] notes about your settings.
Release	What DjangoCloud did to roll your release out: deploying, health checks, and why a release failed.

Searching

• Search shows lines that contain your text. Separate several terms with commas or new lines, up to 10.
• Exclude hides lines that contain any of its terms, which is handy for silencing a noisy health check.
• Time range: the last 15 minutes, 1 hour, 6 hours, 24 hours, 7 days, or all, or a custom range (in UTC).
• Level: all levels, warnings and errors, or errors only. The level is read from the line’s own text.
• Source: app, build or release.
• Live keeps the view up to date every few seconds. Switch it off to read without it moving.

The page shows the newest 300 matching lines and tells you when there are more, so narrow the range or add a search term to get to older ones. Lines keep the time your app logged them.

From the terminal

djangocloud logs                       # the latest 100 lines
djangocloud logs -f                    # keep streaming (Ctrl-C to stop)
djangocloud logs -n 500 --source app   # only your app's output
djangocloud logs --since 2h            # 90s, 30m, 2h or 7d

How long logs are kept

Your app’s log is collected every couple of minutes, so the newest lines can be a little behind. Lines are kept for 3 days on your own AWS account (Starter), 7 days on Company and 30 days on Enterprise, and at most 20,000 lines per deployment, so one very noisy app can’t fill the database. See Known limitations.

Metrics

The Metrics tab shows two charts: CPU and Memory, each as a percentage of what your server size allows. 100% means the server is fully used.

• Ranges: 1 hour, 6 hours, 24 hours, 7 days or 30 days.
• Underneath each chart: the latest value, and the average, peak and low for the range you chose.
• Hover over a chart to read it. A guide line, a dot on the line and a callout show the time (in your own timezone) and the value, so you can see exactly when the load was high. On a touch screen, touch and drag. With a keyboard, focus a chart and use the left and right arrow keys.
• Averages on long ranges. A chart draws at most 160 points. When a range has more samples than that, each point is the average of several, and the callout says so. The Peak under the chart always comes from the real samples, so it can be higher than any point you see.

Samples are taken every 5 minutes while your app is running and kept for 30 days. A new deployment shows “No samples in this time range yet” until the first one arrives.

In the deployments list

Each row of the Deployments list has two small graphs, CPU and Memory, for the last 24 hours. They have the same hover callout. The list refreshes every 10 seconds, and pauses while your pointer is on a graph so the callout doesn’t disappear while you read it.

Metrics from the terminal

djangocloud metrics              # the last hour
djangocloud metrics --since 6h
djangocloud metrics --json       # the samples, for scripts

It draws CPU and memory as small charts with the latest value, the peak and the average for the range. The samples are the same ones the Metrics tab uses: every 5 minutes, kept for 30 days, so --since can reach back at most 30 days.

Status from the terminal

djangocloud status          # is it live and answering, its URL, size and latest releases
djangocloud status --json   # the same as JSON, for scripts