chore: Reduce Docker healthcheck interval #1978
Conversation
In Docker Compose, the helathcheck interval can add an artificial delay because it waits for the first healthcheck to run. Signed-off-by: Charith Ellawala <charith@cerbos.dev>
| @@ -7,7 +7,7 @@ ENV CERBOS_CONFIG="__default__" | |||
| VOLUME ["/policies", "/tmp", "/.cache"] | |||
| ENTRYPOINT ["/cerbos"] | |||
| CMD ["server"] | |||
| HEALTHCHECK --interval=1m --timeout=3s CMD ["/cerbos", "healthcheck"] | |||
| HEALTHCHECK --interval=10s --timeout=2s --retries=2 --start-period=5s CMD ["/cerbos", "healthcheck"] | |||
There was a problem hiding this comment.
I don't think start period < interval makes sense?
start period provides initialization time for containers that need time to bootstrap. Probe failure during that period will not be counted towards the maximum number of retries
If the interval is greater than the start period, then there won't be any probes to fail anyway.
I think we want to set something like --start-interval=1s as well so that we can detect when Cerbos is ready before the first regular probe.
start interval is the time between health checks during the start period.
edit: actually the start interval defaults to 5s so my reasoning here is probably not completely sound, but I still think it makes sense to reduce it since Cerbos typically starts up very quickly.
There was a problem hiding this comment.
Good point about --start-period. The reason I did
7F99
n't include --start-interval is because it was only introduced in Docker Engine 25.0 which was only just released. I think it'd take a little while for people to upgrade.
In Docker Compose, the helathcheck interval can add an artificial delay
because it waits for the first healthcheck to run.
Signed-off-by: Charith Ellawala charith@cerbos.dev