health check add "max_fails" support

[huxos]

Hi,
I use consul-templates to render nginx config file. I found that the nginx is always reload twice in a short interval, as the check state switching.
Would there be any interest in adding max_fails support for health check.

thanks.

[slackpad]

Hi @huxos there have been some requests for more sophisticated health check filtering, can you describe your use case in a little more detail, and how you'd imagine "max_fails" would work?

[huxos]

it means that after several consecutive failures change the check state to critical , not the first time failures discovered.

[slackpad]

Ok thanks. I'm going to mark this as a thinking ticket so we can see if something simple will meet a good 80% cut at people's needs.

[djenriquez]

+1, will satisfy our requirements as an alternative to #1212

[cruatta]

👍

[dbresson]

+1, but also add minimum successes for recovering to healthy.

The lack of this is also the only reason I'm not using Consul Enterprise. Constant notifications because of a random blip in a health check is something I can't deal with.

[morokin]

+1

[Perlovka]

+1

[sirajmansour]

+1 this very important in leader election scenarios where a session could get invalidated just because a service check skipped a beat which would trigger un-necessary change in leader.

[WolverineFan]

To steal a page from a tool that already support this, Amazon ELB health checks have:

    healthy_threshold   = 2
    unhealthy_threshold = 2

This small change helps prevent a tremendous amount of flapping.

[slackpad]

Closing this as a duplicate of #2890.