Health checks

HivePaaS has two kinds of health check:

• the container's, which Docker runs inside each container, to restart one that stopped working, and to know when a new one is ready;
• the app's, which HivePaaS runs from outside against a URL, to tell you when the app is down.

The container's health check

In the app's Container Settings, under the health check, choose its Mode:

• Inherit: the image's own HEALTHCHECK, if it has one;
• CMD: a command run directly, such as curl -f http://localhost:8080/health;
• CMD-SHELL: a command run by the container's shell, such as pg_isready -U "$POSTGRES_USER".

The command succeeds, with exit code 0, while the container is healthy. Then:

Setting	What it does
Interval	The time between two checks.
Timeout	The longest one check may take.
Retries	How many checks in a row must fail before the container is unhealthy.
Start Period	The time a starting container gets before failures count.
Start Interval	The time between checks during the start period.

Docker replaces a container that turns unhealthy. The command runs inside the container, so the tool it uses, such as curl or wget, must be in the image.

The app's health checks

In the app's Periodic Jobs, create a health check:

Setting	What to give
Type	REST for an HTTP endpoint, or GRPC for a gRPC health service.
URL, Method, Body	The request to make, such as GET https://shop.example.com/health.
Return Code Must Be	The status codes that pass, such as 200,204.
Return Body Must Be	Optional: text the body must equal or match, or JSON it must equal or contain.
Scheduling	How often to check, and the timeout, retries and delay between them.
Notification Configuration	Who hears when it fails, and how often at most, with Min Send Interval.

When a check fails, HivePaaS notifies the target you chose. A scheduled job can run when the app's health checks fail, and when they pass again.