Page MenuHomePhabricator

(mw.track) Migrate timing.growthExperiments.* to statslib
Closed, ResolvedPublic

Description

To complete the migration to statslib client code also needs to be updated.

GrowthExperiments client code contains the following timing calls to migrate:

  • /extensions/GrowthExperiments/modules/ext.growthExperiments.DataStore/GrowthTasksApi.js
    • 496 timing.growthExperiments.specialHomepage.growthTasksApi
  • /extensions/GrowthExperiments/modules/ext.growthExperiments.Homepage.Impact/App.vue
    • 80 timing.growthExperiments.newImpact.' + inject( 'RENDER_MODE' ) + '.mounted
  • /extensions/GrowthExperiments/modules/ext.growthExperiments.Homepage.SuggestedEdits/index.js
    • 70 timing.growthExperiments.specialHomepage.modules.suggestedEditsTimeToInteractive
    • 77 timing.growthExperiments.specialHomepage.modules.suggestedEditsTimeToInteractive.serverSideStartInclusive
    • 85 timing.growthExperiments.specialHomepage.modules.suggestedEditsLoadingComplete
  • /extensions/GrowthExperiments/modules/ext.growthExperiments.SuggestedEditSession/index.js
    • 627 timing.growthExperiments.suggestedEdits.taskEditorReady. + this.taskType
    • 646 timing.growthExperiments.suggestedEdits.guidanceShown. + this.taskType
  • /extensions/GrowthExperiments/modules/homepage.init.js
    • 22 timing.growthExperiments.specialHomepage.navigationDuration
    • 29 timing.growthExperiments.specialHomepage.navigationTransferSize
    • 35 timing.growthExperiments.specialHomepage.paintStartTime

Now that T355837 is resolved, this task is actionable. Example of migration: mediawiki/core/+/1105813

Event Timeline

There are a very large number of changes, so older changes are hidden. Show Older Changes
Sgs triaged this task as Medium priority.Dec 11 2024, 4:37 PM
Sgs moved this task from Backlog to Statslib migration on the Growth-Team (Maintenance) board.
Sgs updated the task description. (Show Details)
Sgs raised the priority of this task from Medium to High.Jan 7 2025, 5:05 PM

Change #1109391 had a related patch set uploaded (by Cyndywikime; author: Cyndywikime):

[mediawiki/extensions/GrowthExperiments@master] metrics: Migrate GrowthExperiments client code to statslib

https://gerrit.wikimedia.org/r/1109391

Urbanecm_WMF subscribed.

Moving back to Doing, as the patch has a V-1 and is not mergeable yet.

Sgs renamed this task from Migrate GrowthExperiments client code to statslib to (mw.track) Migrate timing.growthExperiments.* to statslib.Jan 22 2025, 6:05 PM

Change #1109391 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] metrics: Migrate GrowthExperiments client code to statslib

https://gerrit.wikimedia.org/r/1109391

Change #1129303 had a related patch set uploaded (by Sergio Gimeno; author: Sergio Gimeno):

[mediawiki/extensions/GrowthExperiments@master] statslib(Impact): fix invalid label value for render_mode

https://gerrit.wikimedia.org/r/1129303

Change #1129303 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] statslib(Impact): fix invalid label value for render_mode

https://gerrit.wikimedia.org/r/1129303

Change #1131281 had a related patch set uploaded (by Sergio Gimeno; author: Sergio Gimeno):

[mediawiki/extensions/GrowthExperiments@master] fix(stats): fix invalid label value for task_type

https://gerrit.wikimedia.org/r/1131281

Testing notes: Currently - testwiki wmf.22 - the following error is displayed for &gesuggestededit=1&veaction=edit:
Invalid label value for mediawiki_GrowthExperiments_task_editor_ready_seconds task_type "link-recommendation"

Are those logstash reports (https://logstash.wikimedia.org/goto/a5e8693844fc133fe7d698f7cebd7c46) related to the issue?

normalized_message": "Invalid timing value for mediawiki_GrowthExperiments_suggested_edits_server_tti_seconds"

Seems to started around March 21/2025 - https://logstash.wikimedia.org/goto/76e2d3c791caf528cfed723cd47e0440

Change #1131281 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] fix(stats): fix invalid label value for task_type

https://gerrit.wikimedia.org/r/1131281

Change #1131821 had a related patch set uploaded (by Krinkle; author: Krinkle):

[mediawiki/extensions/GrowthExperiments@master] homepage: Add `homepage_transfersize_bytes_bucket_total` metric

https://gerrit.wikimedia.org/r/1131821

Change #1131823 had a related patch set uploaded (by Krinkle; author: Krinkle):

[mediawiki/extensions/GrowthExperiments@master] homepage: rename _duration to _loadeventend

https://gerrit.wikimedia.org/r/1131823

Change #1131823 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] homepage: rename _duration to _loadeventend

https://gerrit.wikimedia.org/r/1131823

Change #1131821 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] homepage: Add `homepage_transfersize_bytes_total` metric

https://gerrit.wikimedia.org/r/1131821

Change #1133157 had a related patch set uploaded (by Michael Große; author: Krinkle):

[mediawiki/extensions/GrowthExperiments@wmf/1.44.0-wmf.22] homepage: Add `homepage_transfersize_bytes_total` metric

https://gerrit.wikimedia.org/r/1133157

Change #1133158 had a related patch set uploaded (by Michael Große; author: Krinkle):

[mediawiki/extensions/GrowthExperiments@wmf/1.44.0-wmf.23] homepage: Add `homepage_transfersize_bytes_total` metric

https://gerrit.wikimedia.org/r/1133158

Change #1133157 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@wmf/1.44.0-wmf.22] homepage: Add `homepage_transfersize_bytes_total` metric

https://gerrit.wikimedia.org/r/1133157

Change #1133158 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@wmf/1.44.0-wmf.23] homepage: Add `homepage_transfersize_bytes_total` metric

https://gerrit.wikimedia.org/r/1133158

Mentioned in SAL (#wikimedia-operations) [2025-04-01T20:29:58Z] <taavi@deploy1003> Started scap sync-world: Backport for [[gerrit:1133157|homepage: Add homepage_transfersize_bytes_total metric (T382003)]], [[gerrit:1133158|homepage: Add homepage_transfersize_bytes_total metric (T382003)]], [[gerrit:1133232|Don't add WikiLove icon to Minerva (T390642)]]

Mentioned in SAL (#wikimedia-operations) [2025-04-01T20:37:19Z] <taavi@deploy1003> migr, taavi: Backport for [[gerrit:1133157|homepage: Add homepage_transfersize_bytes_total metric (T382003)]], [[gerrit:1133158|homepage: Add homepage_transfersize_bytes_total metric (T382003)]], [[gerrit:1133232|Don't add WikiLove icon to Minerva (T390642)]] synced to the testservers (https://wikitech.wikimedia.org/wiki/Mwdebug)

Mentioned in SAL (#wikimedia-operations) [2025-04-01T20:46:57Z] <taavi@deploy1003> Finished scap sync-world: Backport for [[gerrit:1133157|homepage: Add homepage_transfersize_bytes_total metric (T382003)]], [[gerrit:1133158|homepage: Add homepage_transfersize_bytes_total metric (T382003)]], [[gerrit:1133232|Don't add WikiLove icon to Minerva (T390642)]] (duration: 16m 59s)

There are still some error reports (much less though than before) for wmf.23 -https://logstash.wikimedia.org/goto/88305b0939400634c82cf8628c31e62d.

There are still some error reports (much less though than before) for wmf.23 -https://logstash.wikimedia.org/goto/88305b0939400634c82cf8628c31e62d.

I'm not sure what is going on there. The error message is as follows:

Invalid timing value for mediawiki_GrowthExperiments_suggested_edits_server_tti_seconds

But the value it is logging is:

ext.growthExperiments.Homepage.SuggestedEdits/index.js
serverDuration = mw.now() - mw.config.get( 'GEHomepageStartTime' );

Is GEHomepageStartTime sometimes not set to something sensible?

Change #1133998 had a related patch set uploaded (by Michael Große; author: Michael Große):

[mediawiki/extensions/GrowthExperiments@master] Log more data if serverDuration is not a number

https://gerrit.wikimedia.org/r/1133998

I still don't understand what is going on, but added a bit more logging. Maybe it will tell us something useful.

I still don't understand what is going on, but added a bit more logging. Maybe it will tell us something useful.

The error seems to start on March 19/2025 - https://logstash.wikimedia.org/goto/fb0c2951523fbd34ff33e0823ed771c7. The rate of error has been about the same (high) since then

Screen Shot 2025-04-23 at 5.38.22 PM.png (642×2 px, 178 KB)

Change #1133998 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] Log more data if serverDuration is invalid

https://gerrit.wikimedia.org/r/1133998

It seems the additional logging data change didn't work out as expected. It seems the error logger overwrites whatever error.error_context would initially have. I look into it.

Actually what happens is the catch clause is never executed because mw.track returns errors rather than throwing them. Working on a fix.

Change #1142597 had a related patch set uploaded (by Sergio Gimeno; author: Sergio Gimeno):

[mediawiki/extensions/GrowthExperiments@master] fix(logging): temporary subscribe to analytics uncaught errors

https://gerrit.wikimedia.org/r/1142597

Change #1142597 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] fix(logging): temporary subscribe to analytics uncaught errors

https://gerrit.wikimedia.org/r/1142597

This comment was removed by Sgs.

Change #1147720 had a related patch set uploaded (by Sergio Gimeno; author: Sergio Gimeno):

[mediawiki/extensions/GrowthExperiments@master] stats(SuggestedEdits): avoid tracking negative tti durations

https://gerrit.wikimedia.org/r/1147720

Change #1147720 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] stats(SuggestedEdits): avoid tracking negative tti durations

https://gerrit.wikimedia.org/r/1147720