mirror of
https://github.com/wekan/wekan.git
synced 2026-11-20 15:45:01 -05:00
#6768: the recovery scan retried every pending activity notification each round, and one whose delivery fails for good - its activity deleted or changed since, a recipient who lost access, a plan with nothing or too much to send - failed the same way every round, forever, so the list only grew. Such an error is terminal now: after a ten-minute grace (an activity being written is never touched) the scan cancels that notification through the same cancellation an administrator uses, as the system, which compacts it to a cancelled receipt. A pending notification older than ACTIVITY_NOTIFICATION_MAX_AGE_DAYS (7) is cancelled as expired instead of being delivered a week late. Network and other transient failures are still retried. Admin Panel -> Problems gets "Retry all" and "Cancel older..." for the whole list at once, site administrators only. #6769: before each run synced-cron asks whether a job of that name is still running - { name, finishedAt: { $exists: false } } - every minute for every job on every instance, and only { intendedAt, name } and the TTL on startedAt were indexed, so it scanned the job history. { name, finishedAt } is created at startup with WeKan's ensureIndex. The new strings are translated in every locale. Thanks to Raymond-Voogt and xet7 ! Fixes #6768, Fixes #6769.