Improve the failure handling of Data Collection
Categories
(Core :: Privacy: Anti-Tracking, defect)
Tracking
()
People
(Reporter: tjr, Unassigned)
References
(Blocks 1 open bug)
Details
Presently, if any of our data collection efforts fail, we just abort data collection silently. We probably don't even update the pref, so we will continuously try and fail.
The better solution would be when we detect a failure in the data collection, at any point:
a) clear all previously collected metrics
b) set a single metric with a string explaining what part of the collection failed, and if possible, any error message or additional debug information
c) submit that metric only
d) update the pref so we don't try again
Comment 1•17 days ago
|
||
Please help with priority / severity. Is this really a defect or rather a task?
Reporter | ||
Comment 2•17 days ago
|
||
Arguably the fact that things can fail silently without us knowing anything is a problem, so I say defect :)
Reporter | ||
Comment 3•2 days ago
|
||
Add a new string metric, which will be a javascript array. The array will contain single-word strings, indicating what failures we encountered (if any.) We will collect as much data as we generally can.
An irrecoverable error, like "Did not get a Promise back from ContentPageStuff" will also have an entry in errors
- there will just be drastically less data in the telemetry ping. That's okay.
Description
•