Skip to content
This repository has been archived by the owner on Feb 20, 2023. It is now read-only.

Modify startup_timeline_framework_start probe to get it to start working #17972

Closed
mcomella opened this issue Feb 11, 2021 · 3 comments
Closed
Assignees
Labels
eng:qa:not-needed Added by QA to issues that cannot be tested needs:triage Issue needs triage performance Possible performance wins

Comments

@mcomella
Copy link
Contributor

mcomella commented Feb 11, 2021

In mozilla/glam#1141, we determined that having client_id disabled in the ping configuration breaks these probes. That issue will take some time to fix so to get the data in the short term, we can:

  • add client_ids back into the ping
  • move the metrics into the events and metrics pings

The second one seems nice because having to manually transmit our ping adds complexity to the implementation and thus risks adding errors to the results. However, it's more code to change for perhaps no gain if it's implemented correctly. Let's do it if it's simple.

┆Issue is synchronized with this Jira Task

@mcomella mcomella added the performance Possible performance wins label Feb 11, 2021
@mcomella mcomella added this to Needs prioritization in Performance, front-end roadmap via automation Feb 11, 2021
@github-actions github-actions bot added the needs:triage Issue needs triage label Feb 11, 2021
@mcomella mcomella moved this from Needs prioritization to Backlog (prioritized) in Performance, front-end roadmap Feb 17, 2021
@mcomella mcomella self-assigned this Feb 18, 2021
@mcomella mcomella moved this from Backlog (prioritized) to In progress in Performance, front-end roadmap Feb 18, 2021
@mcomella
Copy link
Contributor Author

move the metrics into the events and metrics pings

This isn't possible because the TimingDistribution type doesn't support raw nanos and, without a custom ping, we can't guarantee the timespan type will have been cleared before it's sent.

mcomella added a commit to mcomella/fenix that referenced this issue Feb 18, 2021
…ate docs.

This addresses the root problem we're experiencing for this issue - data
not showing up in GLAM.
mcomella added a commit to mcomella/fenix that referenced this issue Feb 18, 2021
We do this in order to make it easier to analyze in GLAM: see the metric
descriptions for more details.

Additionally, we change the time unit to milliseconds to make it easier
to analyze in GLAM.
mcomella added a commit to mcomella/fenix that referenced this issue Feb 18, 2021
We do this in order to make it easier to analyze in GLAM: see the metric
descriptions for more details.

Additionally, we change the time unit to milliseconds to make it easier
to analyze in GLAM.
mcomella added a commit that referenced this issue Feb 19, 2021
This addresses the root problem we're experiencing for this issue - data
not showing up in GLAM.
mcomella added a commit that referenced this issue Feb 19, 2021
We do this in order to make it easier to analyze in GLAM: see the metric
descriptions for more details.

Additionally, we change the time unit to milliseconds to make it easier
to analyze in GLAM.
@mcomella
Copy link
Contributor Author

Merged in 359f27a

Performance, front-end roadmap automation moved this from In progress to Done Feb 19, 2021
@mcomella mcomella added the eng:qa:not-needed Added by QA to issues that cannot be tested label Feb 19, 2021
@mcomella
Copy link
Contributor Author

move the metrics into the events and metrics pings

This isn't possible because the TimingDistribution type doesn't support raw nanos and, without a custom ping, we can't guarantee the timespan type will have been cleared before it's sent.

I missed this API but it's too late: https://mozilla.github.io/glean/javadoc/glean/mozilla.telemetry.glean.private/-timing-distribution-metric-type/accumulate-samples.html

pkirakosyan pushed a commit to gexsi/user-agent-android that referenced this issue Aug 4, 2021
…ate docs.

This addresses the root problem we're experiencing for this issue - data
not showing up in GLAM.
pkirakosyan pushed a commit to gexsi/user-agent-android that referenced this issue Aug 4, 2021
We do this in order to make it easier to analyze in GLAM: see the metric
descriptions for more details.

Additionally, we change the time unit to milliseconds to make it easier
to analyze in GLAM.
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
eng:qa:not-needed Added by QA to issues that cannot be tested needs:triage Issue needs triage performance Possible performance wins
Development

No branches or pull requests

1 participant