Do not save data for the system users to cuci_user
Closed, ResolvedPublic1 Estimated Story PointsPRODUCTION ERROR
Actions

Description

As detailed in T374994, CheckUser is attempting to include data about the AbuseFilter filter user in the cuci_user table. This fails because the system user name is localised per-wiki and has no global account.

If any system user has no central account, we should ignore this and not send an error to the logging channels. This is because there are many examples of system users without a central account on WMF wikis.

Error

3df49402-c180-4ab2-878e-19a0128fe4ff

message

Unable to find central ID for local user Фільтр зловживань when recording action in cuci_user table.

Details

Related Changes in Gerrit:

Subject	Repo	Branch	Lines +/-
Skip error logs on no central account for system users	mediawiki/extensions/CheckUser	master	+26 -18
Skip error logs on central account for AbuseFilter filter user	mediawiki/extensions/CheckUser	master	+64 -13
Zuul: [mediawiki/extensions/CheckUser] Add AbuseFilter dependency	integration/config	master	+6 -4

Customize query in gerrit

Related Objects
Search...

Status	Subtype	Assigned	Task
			Restricted Task
Resolved		kostajh	T294511 2021 Security Team wikireplicas audit
Declined		None	T284948 Raw IPs of logged-out users disclosed in wiki-replicas
Resolved		Niharika	T324492 Temporary accounts - MVP
Resolved		Madalina	T326816 [Epic] Update features for temporary accounts
Open		None	T355286 [Epic] Globally blocking a temporary account should prevent further account creations
Resolved		Dreamy_Jazz	T368949 [Epic] Implement global autoblocks from global user blocks
Open		None	T384604 [Epic] Use Codex on Special:GlobalContributions
Open		None	T358048 Use Codex HtmlForm in Special:GlobalContributions
Open		None	T358047 Add ToggleButtonGroup to Special:GlobalContributions for displaying results in a table
Open		None	T371917 Use a time period field in Special:GlobalContributions instead of start/end date fields
Resolved		STran	T356292 Return global contributions by temporary accounts given an IP address or range
Resolved		Dreamy_Jazz	T373018 [Epic] Write to, populate, and purge data from the CheckUser central index tables
Resolved	PRODUCTION ERROR	Dreamy_Jazz	T374994 CheckUser is attempting to write to cuci_user for users which have no central account
Resolved	PRODUCTION ERROR	Dreamy_Jazz	T375063 Do not save data for the system users to cuci_user

Event Timeline

Dreamy_Jazz created this task.Sep 18 2024, 8:50 AM

Dreamy_Jazz claimed this task.

Dreamy_Jazz set the point value for this task to 1.

Dreamy_Jazz added a project: AbuseFilter.

Dreamy_Jazz moved this task from Priority Backlog to In Progress on the Trust and Safety Product Sprint (Sprint Beatboxing (Sept 16-27)) board.

Dreamy_Jazz moved this task from Untriaged to Sep 2024 on the Wikimedia-production-error board.

Change #1073741 had a related patch set uploaded (by Dreamy Jazz; author: Dreamy Jazz):

[mediawiki/extensions/CheckUser@master] Skip error logs on central account for AbuseFilter filter user

https://gerrit.wikimedia.org/r/1073741

gerritbot added a project: Patch-For-Review.Sep 18 2024, 9:19 AM

Change #1073743 had a related patch set uploaded (by Dreamy Jazz; author: Dreamy Jazz):

[integration/config@master] Zuul: [mediawiki/extensions/CheckUser] Add AbuseFilter dependency

https://gerrit.wikimedia.org/r/1073743

Dreamy_Jazz moved this task from In Progress to Needs review on the Trust and Safety Product Sprint (Sprint Beatboxing (Sept 16-27)) board.Sep 18 2024, 10:56 AM

Change #1073743 merged by jenkins-bot:

[integration/config@master] Zuul: [mediawiki/extensions/CheckUser] Add AbuseFilter dependency

https://gerrit.wikimedia.org/r/1073743

In T375063#10156308, @gerritbot wrote:

Change #1073741 had a related patch set uploaded (by Dreamy Jazz; author: Dreamy Jazz):

[mediawiki/extensions/CheckUser@master] Skip error logs on central account for AbuseFilter filter user

https://gerrit.wikimedia.org/r/1073741

I suggest excluding all system users with isSystemUser instead of special handling this specific case.

Change #1073741 merged by jenkins-bot:

[mediawiki/extensions/CheckUser@master] Skip error logs on central account for AbuseFilter filter user

https://gerrit.wikimedia.org/r/1073741

Maintenance_bot removed a project: Patch-For-Review.Sep 19 2024, 9:30 AM

ReleaseTaggerBot added a project: MW-1.43-notes (1.43.0-wmf.24; 2024-09-24).Sep 19 2024, 10:00 AM

In T375063#10159283, @JJMC89 wrote:

In T375063#10156308, @gerritbot wrote:

Change #1073741 had a related patch set uploaded (by Dreamy Jazz; author: Dreamy Jazz):

[mediawiki/extensions/CheckUser@master] Skip error logs on central account for AbuseFilter filter user

https://gerrit.wikimedia.org/r/1073741

I suggest excluding all system users with isSystemUser instead of special handling this specific case.

Excluding the error logs for system users in general could work, but I feel like AbuseFilter is a specific exception to the rule of system users sharing the same username (and therefore being able to have a central account).

I can't seem to find the task, but my understanding is that the filter user having a different username for each wiki is a bit of an exception to the normal. For example, the MediaWiki maintenance script system user is not localised.

I'd propose that if other examples appear, we can change the check to just check for a system user. However, otherwise I'd prefer to keep just this limited exception so that we can still see examples of other system users not having central accounts.

This will be hard to test locally and as the QA queue is rather full, I propose that we check that the logs no longer appear once this is on production.

In T375063#10160971, @Dreamy_Jazz wrote:

In T375063#10159283, @JJMC89 wrote:

In T375063#10156308, @gerritbot wrote:

Change #1073741 had a related patch set uploaded (by Dreamy Jazz; author: Dreamy Jazz):

[mediawiki/extensions/CheckUser@master] Skip error logs on central account for AbuseFilter filter user

https://gerrit.wikimedia.org/r/1073741

I suggest excluding all system users with isSystemUser instead of special handling this specific case.

Excluding the error logs for system users in general could work, but I feel like AbuseFilter is a specific exception to the rule of system users sharing the same username (and therefore being able to have a central account).

I can't seem to find the task, but my understanding is that the filter user having a different username for each wiki is a bit of an exception to the normal. For example, the MediaWiki maintenance script system user is not localised.

I'd propose that if other examples appear, we can change the check to just check for a system user. However, otherwise I'd prefer to keep just this limited exception so that we can still see examples of other system users not having central accounts.

There are some more examples now:

mswiki Babel AutoCreate
moswiki Maintenance script
lrcwiki Global rename script
kowiki 바벨 자동만듦

There are many more as well (https://logstash.wikimedia.org/goto/71a12a514aa5a8c94d51c3ceaf170471).

Dreamy_Jazz moved this task from Done to In Progress on the Trust and Safety Product Sprint (Sprint Beatboxing (Sept 16-27)) board.Sep 27 2024, 10:50 AM

Change #1076175 had a related patch set uploaded (by Dreamy Jazz; author: Dreamy Jazz):

[mediawiki/extensions/CheckUser@master] Skip error logs on no central account for system users

https://gerrit.wikimedia.org/r/1076175

gerritbot added a project: Patch-For-Review.Sep 27 2024, 11:17 AM

Thanks. I've uploaded a new patch that will skip the error log for all system users (not just AbuseFilter).

Change #1076175 merged by jenkins-bot:

[mediawiki/extensions/CheckUser@master] Skip error logs on no central account for system users

https://gerrit.wikimedia.org/r/1076175

ReleaseTaggerBot edited projects, added MW-1.43-notes (1.43.0-wmf.25; 2024-10-01); removed MW-1.43-notes (1.43.0-wmf.24; 2024-09-24).Sep 27 2024, 1:00 PM

In T375063#10160984, @Dreamy_Jazz wrote:

This will be hard to test locally and as the QA queue is rather full, I propose that we check that the logs no longer appear once this is on production.

Same as above -- let's check if the message goes away after next week's train.

Maintenance_bot removed a project: Patch-For-Review.Sep 27 2024, 1:30 PM

kostajh mentioned this in T376458: Unable to find central ID for local user {username} when recording action in cuci_user table..Oct 4 2024, 9:57 AM

https://logstash.wikimedia.org/goto/53134cd3698fa55fb2c7f7db98353ac0 shows that we are not getting these errors for system users anymore.

Do not save data for the system users to cuci_userClosed, ResolvedPublic1 Estimated Story PointsPRODUCTION ERRORActions