Shameless plug for a server monitoring macOS app I built. I built that too for the servers I managed myself. Instead of working through a service or a self-hosted monitoring tool that I then over to monitor itself again, I thought running it on my laptop was pretty nice too. One of the next features I've been wanting to build is uptime monitoring.<p><a href="https://kitaso.app/" rel="nofollow">https://kitaso.app/</a><p>Sure there is the downside of having to be at your laptop or having it on, but the upside is that it has very few moving parts and is very simple and it just has one set price.
I'm myself very happy with <a href="https://healthchecks.io/" rel="nofollow">https://healthchecks.io/</a> for this purpose.
I've been toying with a similar idea for ~6 months: A lightweight job status dashboard.<p>I wanted something that required no setup, but could just push success/failure messages to as part of various cron jobs, windows tasks, and shell scripts we run throughout our organization.<p>StatShed server: <a href="https://github.com/statshed/statshed-server" rel="nofollow">https://github.com/statshed/statshed-server</a>
StatShed go-cli: <a href="https://github.com/statshed/statshed-gocli" rel="nofollow">https://github.com/statshed/statshed-gocli</a><p>The idea is kind of like "ntfy.sh", but for jobs status. You can send a "started" message at the beginning, update a "status" message periodically throughout the job, then send a "failed" message (optionally with logs) or a "success". Then a web dashboard gives you an overview with ability to drill down.<p>We use Icinga for monitoring and paging, but this just gives an overview for a quick look at things we don't want heavy duty monitoring on. Like my laptop backups, information about ansible runs across our fleet, etc.
If you need/want a dashboard it's kinda cool. There's a lot of other options that will do something similar, but not via SSH. Crontab can email you directly, no need for a service.<p>You could also just use systemd timers and do: systemctl --failed -t service
You need some sort of external monitor for a missed pulse style alert. Personally I do something similar by collecting a "last success" metric with Prometheus and alerting on it with Grafana if the value is too far in the past. The local system cannot reliably alert if your job does not fail into the alert path or the system is just down.<p>I can see something like this being a great intermediate option to a full monitoring and alerting stack.
You are missing the biggest value offering here - service such as the one linked here (or healthchecks io that I personally use) will let you know that signal didn't arrive even when other things on your end failed (e.g. cron having a hard time sending failure email due to incorrect smtp credentials).
I use those kind of healthchecks literally for everything, especially for backups - they saved my ass many times over.
For things that actually matters to me, I use systemd timers, Prometheus and alert on failed services, but I do get that this might be a bit much for many setups.
You could just wrap cronic around the command and immediately receive the full output per mail when it fails. I don't see how a heartbeat alone without logs would help to identify temporary failures.<p><a href="https://habilis.net/cronic/" rel="nofollow">https://habilis.net/cronic/</a><p>Your service could accomplish something similar if it had such a wrapper to report both success and failures with logs to a remote server. That would take away the need to run a local MTA, while also detecting with the heartbeat whether the job ran at all.
Thanks for the suggestion!<p>There's currently no explicit "fail" ping, as that's not the main use case I care about. It's valid, though, and I've thought about adding it.<p>Being able to pipe logs to curl would also be useful. I'm concerned that people might accidentally send me private/sensitive data, but I'll consider it.
> I don't see how a heartbeat alone without logs would help to identify temporary failures.<p>It does help. Sometimes jobs unexpectedly don't run at all.
The pulse idea is neat. How do you handle jobs where the expected runtime is longer than the heartbeat interval?
Good question! There's no such thing as a "running" task/job at the moment (i.e., a start signal). The closest thing is lengthening the grace period, but that also means that the notification is delayed, so it's not really a solution. This is in the backlog.
Context helps, shows intent, fosters discussion.
People seem to miss the point of the project. It’s not emailing you on failure. It’s emailing on missing reports of scheduled runs. If you rely on the job to report failure, the machine could go down. In that case you won’t get any email on the failed run.
[dead]
[dead]
i have something similar at <a href="https://heartbeats.dorianmarie.com/" rel="nofollow">https://heartbeats.dorianmarie.com/</a>