These 2 probes can give us very useful information about urlbar usage that is one of the main interaction endpoints. But they were given a too small time window. I think these probes should just never expire so we can check them when we make changes to the urlbar interaction. Or if that's not possible, we should bump them enough for us to complete most changes.
Created attachment 8694243 [details] [diff] [review] patch v1 I think the result type probe is something that will be generally always useful to check the awesomebar usage. The index probe is a little less useful, but since we are touching the awesomebar right in these versions is worth bumping it up still for some versions.
Comment on attachment 8694243 [details] [diff] [review] patch v1 Review of attachment 8694243 [details] [diff] [review]: ----------------------------------------------------------------- The renewals are fine. Can you add the alert_emails & bug_numbers fields to these histograms while you're at it, and update the descriptions to point to docs for the meaning of the enum codes (or document inline).
(In reply to Vladan Djeric (:vladan) -- please needinfo! from comment #2) > Can you add the alert_emails Are alerts mandatory? we look at these histograms after urlbar changes, but they are not made to send "alerts". I'm also not sure what they would alert on enumerated histograms... a change in the distribution?
https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Adding_a_new_Telemetry_probe states alert_emails is optional
Created attachment 8694790 [details] [diff] [review] patch v1.1 (no alerts)
I guess for now I will land without the alert_emails property, and then if it ends up being mandatory I can just land a follow-up patch and fix mdn.
(In reply to Marco Bonardo [::mak] from comment #3) > Are alerts mandatory? we look at these histograms after urlbar changes, but > they are not made to send "alerts". > I'm also not sure what they would alert on enumerated histograms... a change > in the distribution? The alerts fire when there is a very significant change in the distribution of the data in the histogram. There are defenses against noise in the monitoring code, so the false positive rate is extremely low. The alert_emails field also helps us identify owners. You can see examples of past alerts here https://groups.google.com/forum/#!forum/mozilla.dev.telemetry-alerts We don't want to miss big changes in the Telemetry data, and realistically most probe authors don't actively monitor their probes. We certainly don't want to end up with probes that collect data about users and never expire, but are not monitored. So I'd like to have the alert_emails fields for these histograms as well. Alternately, we can report the changes via Bugzilla, and you could just tell us which well-monitored component to file the bugs in (put the component name in the the hgram's description field).