The 1.33.0-wmf.24 version of MediaWiki is blocked[0].
The new version is deployed to group0[1], but can proceed no further until these issues are resolved:
* Citoid should only usurp "<ref" / ctrl+shift+k shortcuts if full configured - https://phabricator.wikimedia.org/T219510 * RefreshLinksJob::runForTitle: transaction round 'RefreshLinksJob::run' already started on commons - https://phabricator.wikimedia.org/T220037
The first issue was deemed noncritical for group1 but the second caused a large spike in DBTransactionError rates. The second issue already has a proposed fix but had not yet merged at the closing of the train deployment window.
If both issues are resolved by tomorrow's train window, group1 and group2 will be promoted in succession at that time (barring further issues).
Thank you for your help resolving these issues!
Your humble train toiler, Dan
[0]. https://phabricator.wikimedia.org/T206678 [1]. https://tools.wmflabs.org/versions/
The 1.33.0-wmf.24 version of MediaWiki remains blocked[0], holding at group0 pending resolution of the following issue:
* RefreshLinksJob::runForTitle: transaction round 'RefreshLinksJob::run' already started on commons - https://phabricator.wikimedia.org/T220037
Errors related to the issue were mitigated somewhat by a backported patch to the EventBus extension. However, the 90% drop in DBTransactionError rates still left us at a hefty ~ 800 errors per minute and the root cause is yet unknown.
Thanks to everyone who is helping to resolve this week's issues!
Just a note that since we are well outside today's train deployment window and we refrain from Friday deploys, any fixes for the outstanding issue can be deployed for group0 but promotion of 1.33.0-wmf.24 will be held until Monday.
Kindly, Dan
On Wed, Apr 3, 2019 at 2:29 PM Dan Duvall dduvall@wikimedia.org wrote:
The 1.33.0-wmf.24 version of MediaWiki is blocked[0].
The new version is deployed to group0[1], but can proceed no further until these issues are resolved:
- Citoid should only usurp "<ref" / ctrl+shift+k shortcuts if full
configured - https://phabricator.wikimedia.org/T219510
- RefreshLinksJob::runForTitle: transaction round 'RefreshLinksJob::run'
already started on commons - https://phabricator.wikimedia.org/T220037
The first issue was deemed noncritical for group1 but the second caused a large spike in DBTransactionError rates. The second issue already has a proposed fix but had not yet merged at the closing of the train deployment window.
If both issues are resolved by tomorrow's train window, group1 and group2 will be promoted in succession at that time (barring further issues).
Thank you for your help resolving these issues!
Your humble train toiler, Dan
[0]. https://phabricator.wikimedia.org/T206678 [1]. https://tools.wmflabs.org/versions/
-- Dan Duvall Software Engineer, Release Engineering Wikimedia Foundation http://wikimediafoundation.org
The 1.33.0-wmf.24 version of MediaWiki was successfully promoted to all wikis today. Thanks to everyone who helped fix outstanding issues from last week!
Kindly, Dan
On Thu, Apr 4, 2019 at 2:36 PM Dan Duvall dduvall@wikimedia.org wrote:
The 1.33.0-wmf.24 version of MediaWiki remains blocked[0], holding at group0 pending resolution of the following issue:
- RefreshLinksJob::runForTitle: transaction round 'RefreshLinksJob::run'
already started on commons - https://phabricator.wikimedia.org/T220037
Errors related to the issue were mitigated somewhat by a backported patch to the EventBus extension. However, the 90% drop in DBTransactionError rates still left us at a hefty ~ 800 errors per minute and the root cause is yet unknown.
Thanks to everyone who is helping to resolve this week's issues!
Just a note that since we are well outside today's train deployment window and we refrain from Friday deploys, any fixes for the outstanding issue can be deployed for group0 but promotion of 1.33.0-wmf.24 will be held until Monday.
Kindly, Dan
On Wed, Apr 3, 2019 at 2:29 PM Dan Duvall dduvall@wikimedia.org wrote:
The 1.33.0-wmf.24 version of MediaWiki is blocked[0].
The new version is deployed to group0[1], but can proceed no further until these issues are resolved:
- Citoid should only usurp "<ref" / ctrl+shift+k shortcuts if full
configured - https://phabricator.wikimedia.org/T219510
- RefreshLinksJob::runForTitle: transaction round 'RefreshLinksJob::run'
already started on commons - https://phabricator.wikimedia.org/T220037
The first issue was deemed noncritical for group1 but the second caused a large spike in DBTransactionError rates. The second issue already has a proposed fix but had not yet merged at the closing of the train deployment window.
If both issues are resolved by tomorrow's train window, group1 and group2 will be promoted in succession at that time (barring further issues).
Thank you for your help resolving these issues!
Your humble train toiler, Dan
[0]. https://phabricator.wikimedia.org/T206678 [1]. https://tools.wmflabs.org/versions/
-- Dan Duvall Software Engineer, Release Engineering Wikimedia Foundation http://wikimediafoundation.org
-- Dan Duvall Software Engineer, Release Engineering Wikimedia Foundation http://wikimediafoundation.org
wikitech-l@lists.wikimedia.org