Business Process (Icinga/Nagios) on Prometheus?

[Sebastian Martinez]

Hi there,

I'm moving from Icinga2 to Prometheus monitor solution and we have some questions about it "How to manage or emulate Business Process on Prometheus side".

I'm thinking to try  with some Alerts with functions (on Prometheus) to emulate BP.

Any advise here?.

Thanks in advance.

[Stuart Clark]

On 2020-05-22 17:28, Sebastian Martinez wrote:
> Hi there,
>

> I'm moving from ICINGA2 to Prometheus monitor solution and we have
> some questions about it "How to manage or emulate BUSINESS PROCESS on

> Prometheus side".
> I'm thinking to try with some Alerts with functions (on Prometheus)
> to emulate BP.
>
> Any advise here?.

What sort of business processes?

--
Stuart Clark

[Sebastian Martinez]

Stuart,

At the present time, we are using BP to detect some behaviors between hosts (UP/DOWN) and services (responses like: LOAD, disk usage, etc..).

For example: some cluster with 6 workers (docker swarm), it has 3 Workers Up and 3 Workers Down - This state doesn't represent an issue - but if we have more than a half down, this is an issue.

As I said I am thinking of using Alerts with functions to emulate this.
Please, your opinions are well received.

Ref: https://community.icinga.com/t/module-business-process/83

Regards.

[Ben Kochie]

Yes, that's trivially easy to do in PromQL alerts.

For example: avg(up{job="food"}) < 0.5

This would only alert if more than half the instances are down.

--
You received this message because you are subscribed to the Google Groups "Prometheus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/a35c5811-944c-453c-a073-25f418919ef9%40googlegroups.com.

[Stuart Clark]

On 2020-05-22 17:42, Sebastian Martinez wrote:
> Stuart,
>
> At the present time, we are using BP to detect some behaviors between
> hosts (UP/DOWN) and services (responses like: LOAD, disk usage,
> etc..).
>
> For example: some cluster with 6 workers (docker swarm), it has 3
> Workers Up and 3 Workers Down - This state doesn't represent an issue
> - but if we have more than a half down, this is an issue.
>
> As I said I am thinking of using Alerts with functions to emulate
> this.
> Please, your opinions are well received.
>
> Ref: https://community.icinga.com/t/module-business-process/83
>

Yes that sounds like alerting. You'd have an alert which fires if more
than X (or X%) of workers are down.

https://prometheus.io/docs/prometheus/latest/configuration/alerting_rules/

--
Stuart Clark