Extend (quietly) GIFFT to support labeled_counter -> keyed count and -> categorical histogram mirroring
Categories
(Toolkit :: Telemetry, enhancement, P2)
Tracking
()
Tracking | Status | |
---|---|---|
firefox135 | --- | fixed |
People
(Reporter: chutten, Assigned: chutten)
References
Details
Attachments
(1 file)
There are 16 (2 of which are TELEMETRY_TEST_*
) kind: count, keyed: true
histograms according to $ cat toolkit/components/telemetry/Histograms.json | jq 'to_entries | .[] | {key: .key, kind: .value.kind, keyed: .value.keyed} | select(.kind == "count") | select(.keyed) | .key'
(please do not judge my neophyte jq
use too harshly. I'm learning)
The direct way to collect this shape of information in Glean would be by labeled_counter
. Right now we support keyed scalar and hgram of kind: boolean. We need to add in support for hgrams of kind: count and keyed: true.
Assignee | ||
Comment 1•9 months ago
|
||
Turns out the mapping to histograms with kind: categorical
is straightforward to add while I'm in the neighbourhood.
Assignee | ||
Comment 2•9 months ago
|
||
These are secret menu items. To use them:
- Metric of type: labeled_counter
telemetry_mirror
must haveh#
prefix before the histogram- For categorical:
- Samples must always be 1, or we'll assert
- Labels must be specified in both probes with the same spelling, case, and
order
Tests have been added to t/c/t/tests/python/test_mirrors.py to ensure these
restrictions are adhered to.
Comment 4•9 months ago
|
||
bugherder |
Description
•