Facebook
TwitterAlfonso Uritec
New Pleskian
- Server operating system version
- Windows Server 2019 Standard
- Plesk version and microupdate number
- Version 18.0.81 Update #4v
On Sunday 25 October 2026, when the clocks went back (03:00 CEST -> 02:00 CET), every scheduled task we have with a "cron style" expression of 5 minutes stopped running and did not recover on its own. We lost about 9 hours of maintenance on three sites before we noticed.
Every other Plesk task on the same server kept working normally: daily backups, hourly jobs, Plesk's own maintenance tasks. Those are the ones whose triggers have explicit start times and no <Repetition> element.
Symptoms (Get-ScheduledTaskInfo)
State : Ready <- not Running, nothing was stuck
LastRunTime : 25/10/2026 1:35:35 <- frozen, ~1h25m before the change
NextRunTime : kept moving forward every 5 minutes
NumberOfMissedRuns : increased by 1 every 5 minutes (109 when we found it)
LastTaskResult : 0
So the trigger kept firing on time, but every single run was rejected. No error, no notification e-mail, nothing in the Plesk interface: the task simply looked fine.
The trigger Plesk generates for these tasks:
<CalendarTrigger>
<StartBoundary>2026-10-25T01:00:00</StartBoundary>
<Repetition>
<Interval>PT5M</Interval>
<Duration>P1D</Duration>
</Repetition>
<ScheduleByDay><DaysInterval>1</DaysInterval></ScheduleByDay>
</CalendarTrigger>
The StartBoundary has no time zone and the repetition window is one day long, starting at 01:00 on the very day of the change.
How to reproduce
Create a task with cron style "0,5,10,15,20,25,30,35,40,45,50,55 * * * *" (the expression your own KB recommends for Windows), fetching any URL. Set the server to a time zone with DST and let it cross the autumn change. The task stops about 35 minutes after the StartBoundary of that day and never resumes.
After discovering the issue we disabled and re-enabled the tasks and that fixed the issue; the task runs again immediately and NumberOfMissedRuns goes back to 0.
Could Plesk generate these triggers either with an unlimited repetition duration or with the start boundary synchronized across time zones?
As it is, every 5-minute task on our server silently dies twice a year, and since the task reports State=Ready and LastTaskResult=0 there is nothing to alert on.
Every other Plesk task on the same server kept working normally: daily backups, hourly jobs, Plesk's own maintenance tasks. Those are the ones whose triggers have explicit start times and no <Repetition> element.
Symptoms (Get-ScheduledTaskInfo)
State : Ready <- not Running, nothing was stuck
LastRunTime : 25/10/2026 1:35:35 <- frozen, ~1h25m before the change
NextRunTime : kept moving forward every 5 minutes
NumberOfMissedRuns : increased by 1 every 5 minutes (109 when we found it)
LastTaskResult : 0
So the trigger kept firing on time, but every single run was rejected. No error, no notification e-mail, nothing in the Plesk interface: the task simply looked fine.
The trigger Plesk generates for these tasks:
<CalendarTrigger>
<StartBoundary>2026-10-25T01:00:00</StartBoundary>
<Repetition>
<Interval>PT5M</Interval>
<Duration>P1D</Duration>
</Repetition>
<ScheduleByDay><DaysInterval>1</DaysInterval></ScheduleByDay>
</CalendarTrigger>
The StartBoundary has no time zone and the repetition window is one day long, starting at 01:00 on the very day of the change.
How to reproduce
Create a task with cron style "0,5,10,15,20,25,30,35,40,45,50,55 * * * *" (the expression your own KB recommends for Windows), fetching any URL. Set the server to a time zone with DST and let it cross the autumn change. The task stops about 35 minutes after the StartBoundary of that day and never resumes.
After discovering the issue we disabled and re-enabled the tasks and that fixed the issue; the task runs again immediately and NumberOfMissedRuns goes back to 0.
Could Plesk generate these triggers either with an unlimited repetition duration or with the start boundary synchronized across time zones?
As it is, every 5-minute task on our server silently dies twice a year, and since the task reports State=Ready and LastTaskResult=0 there is nothing to alert on.
