Realized the stat_daily data is not really useful for the chart. There would only be 7 data points for the 1 week chart, and there would be no meaningful information about when up/down events occurred.

Ideally we'd have a stat_hourly covering a couple of weeks. Currently going from minute to day is a 2 orders of magnitude increase, and one might say it's too big anyway. But doing this would also increase a bit of complexity, and I don't know if the data would be useful elsewhere.

Rebased on master, did a quick test and seems to work okay. Still have a couple of issues:

• Will return value 0 even for monitors without ping value. Might need special handling for this
• No handling for maintenance periods

Depends on #4267

#4267 had been merged, for your info.

@louislam have you thought about how to store past maintenance information? I'm leaning towards adding a new column in the stats tables, so that it's indexable. However, if we think about having many other values to keep track of in the future, it may be better to have an "extras" column, then json encode the values. {"maintenance": 10, "slow_response": 2}

Yes, I think it is a good idea to have a column for extra stats.

Maintenance should now also be shown.

Branch will be rebased again once #4406 is merged.

Also fixed a long standing issue where the ping line would get weirdly connected if the server was turned off for a while.

From my point of view, this is mergable. (tested and this works as intended)
@louislam would you like to review this too?

I guess that is a no on an additional review.
=> merging with junior maintainer approval
(if you still have something, I will obv. provide a PR as previously done ^^)