1197823 - Stop sending new data to the "v2" Telemetry backend

Status: RESOLVED FIXED

(Cloud Services Graveyard :: Metrics: Pipeline, defect, P3)

Description

[Mark Reid [:mreid]]

The v2 backend will soon stop aggregating histogram data (see bug 1179751). 

Other than that, there are very few users of the v2 data - the main use being a few remaining scheduled jobs that rely on the v2 "mapreduce" code.

We should update these jobs to use the v4 data, then stop sending new data to the old backend.

Note that we do not want to actively delete the historic raw data from v2. We should simply allow it to expire from S3 based on our usual retention policy.

Work items:
- Update scheduled telemetry jobs to use v4 data
- Update the "tee" server to stop relaying data to the old backend
- Retire the "telemetry-server-stack" Cloudformation stack, if possible (some parts of it may still be needed)
- Implement HTTP Edge spec for v4 pipeline (bug 1137424).

Comment 1

[Mark Reid [:mreid]]

Another thing to note: there are still quite a few old clients who are submitting v2 data, so those will continue to go to the old backend. I'm thinking of "loop" error reports, FxOS first-time-use pings, and older Firefox / Fennec / others.

We could send that data into the new pipeline, but it would require some work (particularly the decoders would need to be updated to handle these older formats).

Comment 2

[Mark Reid [:mreid]]

Everything using the old telemetry pipeline has now been decommissioned, and data has stopped being sent to the old backend.