Audit and improve JobQueue stability and performance (2017)
Closed, ResolvedPublic
Actions

Assigned To

Authored By

	Krinkle
	Sep 14 2017, 8:48 AM

Description

This is a dummy task to track various activities by the Performance Team this year surrounding investigation and optimisation of JobQueue code in MediaWiki and Wikimedia's jobrunner service.

Revisions and Commits

rEWBA extension-Wikibase
	rEWBA081c1cd78f9d Add cause action/agent tracking to link refresh jobs
rMW MediaWiki
	rMW8f829de5f040 Add action/user tracking to link refresh jobs
	rMW2cb965c5a539 Set getDeduplicationInfo() for HTMLCacheUpdateJob
	rMWd0c373fdf2f2 De-duplicate HTMLCacheUpdate jobs with a page array of size 1
	rMWb4ab40d2e89d Enforce $wgReadOnly for job queues
	rMW3daad1264bb8 Push all DeferredUpdates to POSTSEND queue when running that queue
	rMWa549313076b4 Avoid lock acquisition errors for multi-title refreshlinks jobs
	rMW95fdff36c251 Make DeferredUpdates detect LBFactory transaction rounds
	rMW24842cfac0f3 Use AutoCommitUpdate instead of Database->onTransactionIdle
	rMWd80fca05e18d Better handling of jobs execution in post-connection shutdown
	rMWcb7c910ba72b Fix old regression in HTMLCacheUpdate de-duplication
	rMW70d1bc00919e Make workItemCount() smarter for htmlCacheUpdate/refreshLinks
	rMW6898d06f8fec Disable rebound CDN purges for backlinks in HTMLCacheUpdateJob
	rMW9ae936b197ea Prefer that enqueueable post-send updates are run immediately

Related Objects
Search...

Status	Subtype	Assigned	Task
Resolved	Release	thcipriani	T170632 1.30.0-wmf.14 deployment blockers
Resolved		aaron	T175897 Audit and improve JobQueue stability and performance (2017)
Resolved		Ladsgroup	T173710 Job queue is increasing non-stop
Resolved		daniel	T174422 Make dbBatchSize in WikiPageUpdater configurable
Open		None	T178804 When processing changes to Wikibase SiteLinks on the client, only trigger updates for sitelinks that are actually shown in the sidebar.
Resolved		Addshore	T178806 Wikibase: Batch HTMLCacheUpdateJobs across changes
Declined		None	T178810 Wikibase: Increase batch size for HTMLCacheUpdateJobs triggered by repo changes.
Resolved		aaron	T124418 Investigate massive increase in htmlCacheUpdate jobs in Dec/Jan
Resolved		daniel	T164173 Cache invalidations coming from the JobQueue are causing lag on several wikis
Resolved	PRODUCTION ERROR	None	T171370 ERROR: "LBFactory::getEmptyTransactionTicket: WikiPageUpdater::injectRCRecords does not have outer scope"
Resolved	PRODUCTION ERROR	daniel	T172320 Error in Wikibase/client/includes/Changes/InjectRCRecordsJob.php line 120: Bad value for parameter $params: $params['change'] not set.
Resolved		daniel	T172394 InjectRCRecordsJob: include the serialize EntityChange object in the job spec
Resolved		aaron	T130795 Make JobQueueGroup/JobQueue error out on write ops in read-only mode
Resolved		aaron	T168723 LinksUpdate totally broken when JobQueueDB is in use
Resolved	PRODUCTION ERROR	aaron	T173462 LinksUpdate::acquirePageLock: Cannot flush pre-lock snapshot because writes are pending
Resolved		Jrbranaa	T173477 wmf.14 Blocker - Post Mortem - Cannot flush pre-lock snapshot because writes are pending

Event Timeline

Krinkle created this task.Sep 14 2017, 8:48 AM

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptSep 14 2017, 8:48 AM

Krinkle triaged this task as Low priority.Sep 14 2017, 8:48 AM

Krinkle added a project: Performance-Team.

Krinkle added a commit: rMW9ae936b197ea: Prefer that enqueueable post-send updates are run immediately.

Krinkle added commits: rMW6898d06f8fec: Disable rebound CDN purges for backlinks in HTMLCacheUpdateJob, rMW70d1bc00919e: Make workItemCount() smarter for htmlCacheUpdate/refreshLinks, rMWcb7c910ba72b: Fix old regression in HTMLCacheUpdate de-duplication.

Krinkle added a subscriber: aaron.

Krinkle removed a subscriber: Performance-Team.

Krinkle added subtasks: T173710: Job queue is increasing non-stop, T124418: Investigate massive increase in htmlCacheUpdate jobs in Dec/Jan.Sep 14 2017, 8:50 AM

Krinkle renamed this task from Audit and improve JobQueue performance (2017) to Audit and improve JobQueue stability and performance (2017).Sep 14 2017, 8:53 AM

Krinkle added commits: rMWd80fca05e18d: Better handling of jobs execution in post-connection shutdown, rMW24842cfac0f3: Use AutoCommitUpdate instead of Database->onTransactionIdle, rMW95fdff36c251: Make DeferredUpdates detect LBFactory transaction rounds, rMWa549313076b4: Avoid lock acquisition errors for multi-title refreshlinks jobs, rMW3daad1264bb8: Push all DeferredUpdates to POSTSEND queue when running that queue, rMWb4ab40d2e89d: Enforce $wgReadOnly for job queues.

Krinkle added subtasks: T164173: Cache invalidations coming from the JobQueue are causing lag on several wikis, T130795: Make JobQueueGroup/JobQueue error out on write ops in read-only mode, T168723: LinksUpdate totally broken when JobQueueDB is in use, T173462: LinksUpdate::acquirePageLock: Cannot flush pre-lock snapshot because writes are pending.

Krinkle updated the task description. (Show Details)

Krinkle moved this task from Inbox, needs triage to Backlog: Maintenance, non-prioritized on the Performance-Team board.

Krinkle added a commit: rMWd0c373fdf2f2: De-duplicate HTMLCacheUpdate jobs with a page array of size 1.Sep 14 2017, 12:44 PM

Krinkle added a commit: rMW2cb965c5a539: Set getDeduplicationInfo() for HTMLCacheUpdateJob.Oct 20 2017, 5:52 AM

Krinkle added a commit: rMW8f829de5f040: Add action/user tracking to link refresh jobs.Nov 8 2017, 9:01 AM

Krinkle added a commit: rEWBA081c1cd78f9d: Add cause action/agent tracking to link refresh jobs.Nov 8 2017, 11:04 AM

Krinkle closed this task as Resolved.Jan 18 2018, 10:37 AM

Krinkle assigned this task to aaron.

Ladsgroup closed subtask T173710: Job queue is increasing non-stop as Resolved.Feb 13 2018, 1:14 PM

Krinkle added a project: Wikimedia-Performance-publish.Mar 18 2018, 1:16 AM