Add a edit attempt identifier to the Wikistories contributor data stream
Closed, ResolvedPublic
Actions

Assigned To

Authored By

	nshahquinn-wmf
	Sep 15 2022, 11:22 PM

Description

The wikistories_contribution_event stream captures a variety of events from the Wikistories editing funnel: story_builder_open, add_frame, publish_success, and so on.

However, the stream doesn't have a unique identifier for each edit attempt. Without this, we cannot properly group events from a single journey through the funnel.

Let's add such an ID (call it contribution_attempt_id?). I would suggest using mw.user.generateRandomSessionId; there are existing IDs we could reuse (like the pageview token), but since we have no interest in joining this stream with another one, we should just be explicit about that and create a new one.

Details

Other Assignee: nshahquinn-wmf

Subject	Repo	Branch	Lines +/-
Retain hashed Wikistories contribution_attempt_id	analytics/refinery	master	+1 -0
Add contribution_attempt_id to contribution events	mediawiki/extensions/Wikistories	master	+5 -2
Add Wikistories contribution_attempt_id	schemas/event/secondary	master	+319 -4

Customize query in gerrit

Related Objects

Mentioned In: T321960: Presto returns incorrect data for an added field
T312262: Check up on Wikistories instrumentation

Event Timeline

nshahquinn-wmf created this task.Sep 15 2022, 11:22 PM

nshahquinn-wmf mentioned this in T312262: Check up on Wikistories instrumentation.

nshahquinn-wmf moved this task from Triage to Tracking on the Product-Analytics board.Sep 16 2022, 12:40 AM

@nshahquinn-wmf what is the importance/urgency of this task?

Would you be able to handle the schema part while I make the code change?

In T317934#8264760, @SBisson wrote:

@nshahquinn-wmf what is the importance/urgency of this task?

Low. This will be very useful when we try to use this stream to understand use of the story builder, but I think it will be a fairly long time until we get to that stage. I proposed it in kanban since it seems like we might as well get it done now, but if you want to defer it, I'm fine with that.

Would you be able to handle the schema part while I make the code change?

Yep, no problem. I'll go ahead and do that, and you can merge the patch and change the code whenever you like.

In T317934#8267003, @nshahquinn-wmf wrote:

[...]
Yep, no problem. I'll go ahead and do that, and you can merge the patch and change the code whenever you like.

Sounds like a plan

Change 836266 had a related patch set uploaded (by Neil P. Quinn-WMF; author: Neil P. Quinn-WMF):

[schemas/event/secondary@master] Add Wikistories contribution_attempt_id

https://gerrit.wikimedia.org/r/836266

gerritbot added a project: Patch-For-Review.Sep 28 2022, 6:45 PM

Okay, the patch is ready.

Note that, in addition to adding the contribution_attempt_id field, we will need to update the $schema field to '/analytics/mediawiki/wikistories_contribution_event/1.1.0'.

@SBisson actually, please go ahead and merge the patch as soon as you have a chance (you can still do the instrumentation change whenever). I've been working on improving the description fields in the schema and it will simplify things if I can base it on this patch.

Change 836266 merged by jenkins-bot:

[schemas/event/secondary@master] Add Wikistories contribution_attempt_id

https://gerrit.wikimedia.org/r/836266

Maintenance_bot removed a project: Patch-For-Review.Sep 29 2022, 6:30 PM

Change 841953 had a related patch set uploaded (by Sbisson; author: Sbisson):

[mediawiki/extensions/Wikistories@master] Add contribution_attempt_id to contribution events

https://gerrit.wikimedia.org/r/841953

gerritbot added a project: Patch-For-Review.Oct 12 2022, 5:12 PM

SBisson moved this task from Dev to Code Review on the Inuka-Team (Kanban) board.Oct 12 2022, 5:12 PM

Change 841953 merged by jenkins-bot:

[mediawiki/extensions/Wikistories@master] Add contribution_attempt_id to contribution events

https://gerrit.wikimedia.org/r/841953

ReleaseTaggerBot added a project: MW-1.40-notes (1.40.0-wmf.6; 2022-10-17).Oct 13 2022, 1:00 PM

Maintenance_bot removed a project: Patch-For-Review.Oct 13 2022, 1:29 PM

SBisson moved this task from Code Review to QA on the Inuka-Team (Kanban) board.Oct 14 2022, 1:49 PM

Hi @nshahquinn-wmf, are you satisfied with the changes made?

This has been added. I will move this for sign off.

Sorry for the delay!

I just checked the values of this field that we've received so far. The results are very strange! Since this data goes back 90 days, all the nulls are expected, but the many values of codfw and eqiad are definitely wrong. They should have failed validation against the schema since they don't match the specified pattern (^[0-9a-z]{20}$), but there haven't been any validation errors for this schema.

value	frequency
null	1340
codfw	1215
eqiad	50
7d4a499adcd850ec5769	3
c8d489b1613f97a9124c	3
ce00b7c2a11114e5ea50	2
92de0c331e9d2d29fae3	2
cdfce631ceeaab2e262c	2
03f21bdc208d425b7ffd	2
88f3a3f6b6b005c520e8	2
...	...

However, if I only look at the data going back to 2022-10-17 (the start of the train where our instrumentation code was deployed), those weird values disappear and everything looks correct.

value	frequency
null	27
7d4a499adcd850ec5769	3
c8d489b1613f97a9124c	3
3bd60ed9a56b8fb2fda0	2
cdfce631ceeaab2e262c	2
65de010762602c8ea53c	2
56e1fecbdd705f7ba006	2
a6c128c4b0315732fbf5	2
92de0c331e9d2d29fae3	2
3aa0bd519016f1a5ed58	2
...	...

So my guess is that it's some problem with the data ingestion or storage. I'll report that to Data Engineering in case it's a sign of a deeper problem, but from our perspective I think this is done.

nshahquinn-wmf mentioned this in T321960: Presto returns incorrect data for an added field.Oct 28 2022, 10:53 PM