Live data from Hacker News

Ask HN: Job Scheduling as a service

news.ycombinator.com

31–40 of 42 posts

Re: Ask HN: Job Scheduling as a service

#31
post #9

Earlier quoted context omitted.

Jenkins has scheduled jobs and an UI. It can also be accessed via API. https://jenkins.io/

This is not with respect to devops or build . My app has schedule functionality for example a scheduled report , so if it where like a webhook call then i dont want to waste the cpu cycle ,now i am using quartz scheduler in my java app .

Jenkins isn't just for devops or build.

Re: Ask HN: Job Scheduling as a service

#32
post #29

Nomad by Hashicorp. You have to do just a little legwork to get it to call an API - just schedule a job that calls the appropriate curl command. When I'm explaining it to people, I explain it is a "distributed cron". You schedule a job either using the command line or via an API to the Nomad server and then it runs the job on any box that has the Nomad agent running and registered to the server. If you combine Nomad…

Nomad is probably a bit overkill for cron. It's a whole k8s alternative isn't it?

It can be. But if you just need a distributed cron, you can just use the raw_exec or exec driver type and run a shell command.

Re: Ask HN: Job Scheduling as a service

#38
I'm surprised no one's mentioned Chronos - it takes a butt load of setup because it runs on top of Mesos (which requires Zookeeper) etc - but if you already have these tools in your stack it's a pretty tidy clustered scheduler: https://medium.com/airbnb-engineering/chronos-a-replacement-...

Edit: if you haven't got this stack around though i'd second calls to use a build tool like Jenkins or Rundeck - of course you probably don't want this on the same Jenkins that builds your jobs... so you'd have to set all that up

Re: Ask HN: Job Scheduling as a service

#39

There are a few online tools, such as: https://www.easycron.com/ https://www.setcronjob.com/ etc. You might be as well writing your own though, since it shouldn't be a difficult task.

Even if you want clustered level resilience? Loads of places i've worked end up running a scheduler on 'just one server' - that ain't redundant brother!
Post reply on HN