Cachet: Feature request: per-hour/12hrs/month/week metrics, like "99.95% uptime for the last month"

[meonkeys]

SLAs generally require visibility into average metrics per some time period. For example: "uptime percentage per month". It would be helpful if Cachet would support reporting metrics in this fashion. Example with "uptime percentage per month" numbers: https://status.heroku.com/uptime .

[abilto]

In order to track this would we need to store the operational status with each incident? If we wanted to track the existing incident status (investigating, watching, etc) we could do a SQL query something like this:

SELECT 
    incidents.created_at AS start_datetime, 
    nextIncident.created_at AS end_datetime, 
    incidents.`status` AS `component_status`, 
    incidents.`component_id` 
FROM 
    incidents 
    INNER JOIN incidents nextIncident 
        ON nextIncident.id > incidents.`id` 
        AND incidents.component_id = nextIncident.component_id 
        AND incidents.status != nextIncident.status 
GROUP BY component_id, start_datetime

If the component status was stored there we could track that over time as well.

[jbrooksuk]

I don't think that this can be efficiently done until https://github.com/CachetHQ/Cachet/pull/1595 is merged in?

[mjrussell]

This would be a great feature. I saw that #1595 is merged, is there any movement on this now?

[jbrooksuk]

This would be a great feature. I saw that #1595 is merged, is there any movement on this now?

It's merged but only into the v3 branch, we're still working on v2.4 first :)

[alexcurtin]

any update on this capability?

[slava-basko]

up