Clarification on modalities of healthcheck.io option

[Marcool04]

Hi there.

First of all, thank you for the very helpful and nicely designed tool that is ddns-updater.

I have been looking into using healthcheck.io, which I self-host a version of on another machine, to monitor the running of ddns-updater. Healthcheck seems to all be setup OK. I can curl the healthcheck endpoint and it gets updated, then if I append a number to the call I can get it into a failed state and am notified.

On the ddns-updater side, I run an up to date version (2.9.0), "bare bone", ie not via docker, on Arch linux, using the AUR package which I maintain https://aur.archlinux.org/packages/ddns-updater. I have the two env variables setup HEALTH_HEALTHCHECKSIO_BASE_URL and HEALTH_HEALTHCHECKSIO_UUID.

However, so far, healthcheck is getting nothing from ddns-updater, and I wonder if this is perhaps by design, and so I have questions about the design.

From reading the codebase, it looks to me like ddns-updater makes a call to healthcheck here:

ddns-updater/internal/update/service.go

Line 373 in b2a446b

err := s.hioClient.Ping(ctx, healthchecksIOState)

which is in the updateNecessary function. Again, if I'm reading this correctly, it means that the healthcheck endpoint only gets called when an update is needed? Correct?

If so, then that is not how I understand the basic idea of healthcheck.io, which requires that HEALTHY services systematically and periodically report that they are healthy. What if ddns-updater last had a successful run, but then crashes silently, or hangs for example? I will get no more calls to healthcheck, and will continue to assume that it is in an "OK" state. In the mental model I have of this setup, the healthcheck API should get called on every single run (I'm not sure I understand the code entirely, but possibly on every tick on the "ticker"

ddns-updater/internal/update/service.go

Line 404 in b2a446b

ticker := time.NewTicker(s.period)

) so that ddns-updater can tell it "I'm still here". Then it's up to healthcheck to keep a watch on time, and if it doesn't hear from ddns-updater in say, 1h, or 1 day (configurable) or whatever, assume that ddns-updater is faulty, and send out a notification as requested in healthcheck.

When is ddns-updater supposed to call healthcheck at present? Am I misreading something here? And if I'm not, what do you think of this proposed change? Would that not be a more logical way to set things up ?

Thanks in advance for your time and thoughts,

Best regards,

Mark.

[Marcool04]

Ping, anybody have thoughts/info on this please?

[Marcool04]

and again ??

[Jeppedy]

I'd love to hear the response, too.