Monitoring: statistics, notifications and checks

Statistics

At the top of the Monitoring tab, four charts show the resource usage for the entire project: RAM, CPU, storage and processes. The readings are taken every 5 minutes. Use the buttons to select the period: 24 hours, 7 days or 30 days. Below each chart, you’ll find the average (Ø) and peak values; for RAM and CPU, the percentage share of each container is also shown.

If your usage consistently reaches the limits of your plan, upgrading to a higher-tier plan or adding extra resources will give you some breathing space (Plan and options). The overview shows processes that were terminated due to a lack of memory.

Notifications

Independently of the Monitoring option, we send an email to the address associated with your customer account if …

• a container fails - due to an error, a restart loop or an unexpected termination - and when it is up and running again,
• an application reports an error - an address responds with a 5xx status or not at all,
• resources become scarce - memory, CPU, storage or processes reach one of your warning thresholds (selectable from 50 to 95 %, default 80 and 95 %). We report each threshold once; you can switch the warning off at your own risk (Resource limits),
• a backup is missing - a backup is scheduled but none has been performed for 30 hours.

Enable the notifications you require and click Save. The settings take effect from the next check run. We will only report a problem once it has persisted for at least 5 minutes, and will send a further notification to confirm that it has been resolved.

Checks using the Monitoring option

With the Monitoring option, we check your services regularly and notify you by email in the event of a fault. On request, we will automatically restart the affected container - up to three times per hour.

• Basic - up to 10 checks of web addresses and TCP ports, every 60 seconds or less frequently.
• Advanced - up to 50 checks starting from 30 seconds, plus checks for containers and resources.

To create a check:

1. Click on Create check and select the type: web address, TCP port, container or resources.
2. Enter a name and select the target. For a web address, specify the path, the expected status (2xx for any successful response or a fixed code such as 200) and, optionally, a keyword that must appear on the page.
3. Select the interval and specify how many consecutive failures will trigger an alert.
4. Decide whether you wish to be notified by email and whether the container should restart automatically.

Each check shows its status. Click the clock icon to open the history, which shows faults, all-clear notifications and automatic restarts. If the limit of three restarts per hour is reached, this will also be shown in the history - in which case the container requires your attention (Finding and fixing errors). Use Pause to suspend a check, for example during maintenance.