You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Data is going to be recorded as a series of sparse records where x=timestamp and y=0 or 1 for if the request was successful.
That can be plotted as a regular line plot with a marker on each entry. You'll need to partition on the name of the CanaryHTTPMonitor and on the Node Name / DaemonSet Pod that the metric was recorded from for each url so they all get their own line.
The entries will be slightly irregularly spaced, and also a binary 0 or 1 will be very spiky if we are having issues.
In order to plot it smoothed we'll need to aggregate over a sliding time window it so that becomes an availability % over time.
Implement this dashboard using https://github.com/weaveworks/grafanalib and generate the json from it locally and commit it into the repo. We'll work out CI for the in the longer term but I can't see the dashboard changing that much once we get these plots working.
Graph of raw data as line plot.
Graph of time window averaged data as a % line plot.
Graph of how long each request took as a line plot.
The text was updated successfully, but these errors were encountered:
Data is going to be recorded as a series of sparse records where
x=timestamp
andy=0 or 1
for if the request was successful.That can be plotted as a regular line plot with a marker on each entry. You'll need to partition on the name of the
CanaryHTTPMonitor
and on theNode Name / DaemonSet Pod
that the metric was recorded from for each url so they all get their own line.The entries will be slightly irregularly spaced, and also a binary 0 or 1 will be very spiky if we are having issues.
In order to plot it smoothed we'll need to aggregate over a sliding time window it so that becomes an availability % over time.
Implement this dashboard using https://github.com/weaveworks/grafanalib and generate the json from it locally and commit it into the repo. We'll work out CI for the in the longer term but I can't see the dashboard changing that much once we get these plots working.
The text was updated successfully, but these errors were encountered: