Probe time series data

[MartinKolarik]

A parent task to track all possible features based on time series data (order roughly based on priority):

1. Logs. Already live, but moving the data away from redis will enable longer retention.
2. Status/uptime. Fairly easy and useful: Credits/Uptime tab #82
   2.1. Better notifications: Better probe downtime tracking and email notifications #27
3. Usage statistics: Tests per probe chart #24
4. CPU/mem statistics.

[jimaek]

Do we need to store logs outside of redis? Lots of data for no good reason

[MartinKolarik]

Well, it's the opposite; they are not nearly important enough to be stored in memory when they can be stored on an SSD. The number of them will still be limited, but we can store a bit more than now. The amount of data will still be negligible compared to measurement results. Possibly later, we can also combine it with the status charts, e.g., allow you to jump to a specific time when something happened.

[jimaek]

Are we talking about the same logs? Probe logs that basically only ever say "measurement Id started" why would you want to jump there

[MartinKolarik]

To see why the probe went down at that time, or generally why it behaved the way it did, that's why we're collecting them in the first place.

[jimaek]

I still don't see the point, that's lots of data that nobody will ever use. People check them rarely and redis fits that