Submit tagged debug pings to standard telemetry endpoint
Categories
(Data Platform and Tools :: Glean: SDK, task, P1)
Tracking
(Not tracked)
People
(Reporter: akomar, Assigned: travis_)
Details
Attachments
(1 external file)
Per [1], tagged debug pings are submitted to staging GCP endpoint. At the time when this was implemented we were concerned about potential delays of forwarding data from AWS - this should no longer be an issue because of changes in infrastructure.
Staging GCP projects will be reorganized this week and no longer suitable for processing debug pings, therefore Glean client should switch to sending them to standard telemetry endpoint [2].
We don't expect any issues related to the switch, but just to be safe I'd advise to have a configurable, alternative endpoint for debug pings. I have tested and confirmed that debug pings sent to AWS endpoint are surfaced in DebugView.
[1] https://bugzilla.mozilla.org/show_bug.cgi?id=1522428
[2] https://incoming.telemetry.mozilla.org
| Reporter | ||
Comment 1•7 years ago
|
||
Cutoff for changes to Android-Components (where this would land) is today. According to Travis, even if we landed this today we wouldn’t get to see it live in Fenix before it was too late to revert.
:whd - two questions:
- How hard would it be for us to live with Glean submitting debug pings to staging endpoint?
- Is AWS Tee currently stable and reliable?
| Reporter | ||
Comment 2•7 years ago
|
||
If the answer to second question above is "yes", I'm thinking that maybe we should just get rid of alternative GCP endpoint and send everything to standard one, i.e. forget about configurable, alternative endpoint for debug pings - that would minimize the risk of introducing bugs there (Travis - would you agree?).
| Assignee | ||
Comment 3•7 years ago
|
||
I would rather have a single endpoint and remove the redirection entirely, if at all possible. Then the selection of the endpoint is the responsibility of configuration and we don't have to do any redirection at all.
Comment 4•7 years ago
|
||
(In reply to Arkadiusz Komarzewski [:akomar] from comment #1)
:whd - two questions:
- How hard would it be for us to live with Glean submitting debug pings to staging endpoint?
Not hard.
- Is AWS Tee currently stable and reliable?
Yes.
| Reporter | ||
Comment 5•7 years ago
|
||
(In reply to Travis Long from comment #3)
I would rather have a single endpoint and remove the redirection entirely, if at all possible. Then the selection of the endpoint is the responsibility of configuration and we don't have to do any redirection at all.
Then I'd say we should do this as soon as is safely possible. If we're not confident with landing this now, let's go with the next release and we'll keep the endpoint running.
Comment 6•7 years ago
|
||
(In reply to Arkadiusz Komarzewski [:akomar] from comment #5)
(In reply to Travis Long from comment #3)
I would rather have a single endpoint and remove the redirection entirely, if at all possible. Then the selection of the endpoint is the responsibility of configuration and we don't have to do any redirection at all.
Then I'd say we should do this as soon as is safely possible. If we're not confident with landing this now, let's go with the next release and we'll keep the endpoint running.
Given that Glean SDK code is frozen as of today, I think this should happen after Fenix launch.
| Reporter | ||
Comment 7•7 years ago
|
||
(In reply to Alessio Placitelli [:Dexter] from comment #6)
Given that Glean SDK code is frozen as of today, I think this should happen after Fenix launch.
Yep, sounds good as long as we land this with the next SDK release.
| Assignee | ||
Comment 8•7 years ago
|
||
Attaching a PR that removes the redirection of tagged pings to GCP and relies upon the configured telemetry endpoint.
| Assignee | ||
Updated•7 years ago
|
| Assignee | ||
Comment 9•7 years ago
•
|
||
Filed Bug #1561269 as follow-up to perform the same fix for glean-core.
| Assignee | ||
Comment 10•7 years ago
|
||
Merged after double checking with Arkadiusz
Description
•