Page MenuHomePhabricator

CX2: Translation unit saving failure should not cause long save retry loop
Closed, ResolvedPublic

Description

Basically porting the CX1 fixes done as part of T163344: Do a root-cause analysis on CX outage during dc switch and get it back online to CX2. Especially, changes from https://gerrit.wikimedia.org/r/#/c/350200/:

Improve CX draft saving logic

  • Increase retry delay after each failure
  • Limit automatic retry attempts to 5, after which user action is needed to attempt saving again.
  • If a save has failed, prevent other save requests other than the the retry attempt
  • Throttle auto-saving: before it would save when a three second pause is detected. This could lead the worst case of a save happening every three seconds. Changed this to use throttle instead, which saves every 15 seconds (taking a very conservative value here on purpose) regardless whether there is a pause in the writing or not.

Event Timeline

santhosh triaged this task as Medium priority.
Nikerabbit renamed this task from CX2: Translaiton unit saving failure should not cause long save retry loop to CX2: Translation unit saving failure should not cause long save retry loop.Jun 14 2017, 6:10 AM

Change 358931 had a related patch set uploaded (by Santhosh; owner: Santhosh):
[mediawiki/extensions/ContentTranslation@master] CX2: Improve the retry mechanism for save failures

https://gerrit.wikimedia.org/r/358931

Change 358931 merged by jenkins-bot:
[mediawiki/extensions/ContentTranslation@master] CX2: Improve the retry mechanism for save failures

https://gerrit.wikimedia.org/r/358931

Nikerabbit moved this task from QA to Done on the Language-2017-July-Sept board.