I am trying to visualize when my web backend is unhealthy using the varnish_backend_up metrics, which take values of 0 for unhealthy and 1 for healthy.
The problem with time series visualization arises when I display wider ranges, such as 7 days. Even if all the backends are unhealthy, it doesn’t show the maximum value. However, when I zoom in, it starts to show that the metric reaches the maximum value.
For boolean values, it is crucial to know when the metrics were 1 or 0. I am not interested in averaging the values over a wider range.
I also tried to use Status history chart but for 7 days range I get:
Too many points to visualize properly.
Update the query to return fewer points.
(2017 points received)
Is there any best practice for visualizing boolean statuses where discrete values are very important?
Wide range:
Zoomed in:
Queries I use:
count(varnish_backend_up{backend=~"web.*"})-sum(varnish_backend_up{backend=~"web.*"})
and for total:
count(varnish_backend_up{backend=~"web.*"})