Hi folks,

Brand new issue. Latest updates to Reflect on Windows 10. First time for this problem on 12/1.

I have a plan that is scheduled to run at 5:00 AM on the 1st as Full, then weekly Differentials and daily Incrementals. It did run as expected, BUT it ran again at 6:00 AM creating two full backups an hour apart. Has never done this before.

Task Manager looks correct. Scheduled backups in Reflect look correct. I do not see anything obvious in the vss logs.

What do you think?

~Bob

Do you happen to live in an area that switched between Daylight Savings and Standard Time this past weekend?  Or more precisely, is your system set to a time zone where that would have occurred?  If not, open Windows Task Scheduler, find the scheduled task that corresponds to running Full backups for this job, select it, and check the History tab.  If it's marked as disabled, which it is on default Windows installations, then there won't be anything useful there.  But if not, it may take a while to populate, but after it does, is there anything useful in there?

Thanks for the reply.

Interesting thought on time change. It occurred on November 4th here, but this IS the first full backup since then. You think that could have caused it?

History is disabled in Task Manager. Reflect logs do show both being run, but nothing there that showed a failure to cause it to run again. Both fulls are the same size, both ran the retention rules.

~Bob

There are a few threads here and in other forums that have nothing to do with Reflect about scheduled tasks behaving oddly after a time change.  In some cases, the anomalous behavior manifests as an unexpected additional occurrence an hour after the normal occurrence, after which the scheduled task continues as normal.  In other cases, the scheduled task completely fails to run and throws some error code, and that persists until the scheduled task is deleted and recreated (or perhaps modifying it is enough, I can't remember).  Both appear to be bugs in Windows Task Scheduler that have existed for many years and persist into the latest versions of Windows -- unless the recent Win10 1809 has finally solved this, although I'm not holding my breath.  The fact that this was the first time that particular scheduled task was executed since your time change is almost certainly NOT a coincidence, and since your behavior is the "unexpected extra occurrence" variation, things should carry on as normal going forward without you having to do anything.

Side note: The other behavior I described, where the scheduled task completely fails, is a perfect example of why I strongly recommend that people who use email notifications enable them for both failures AND successes.  Many people only enable them for failures, but the problem with that strategy is that you won't know if LACK of emails means that everything is going fine or nothing is happening at all.  In this particular bug, since the error in question occurs at a Windows Task Scheduler level, Reflect never even launches, which means that it would never generate a failure email notification about the problem.  On a server that is typically unattended (i.e. people aren't logging onto it on a regular basis), if the administrator has configured Reflect to only send failure notifications, this bug could cause that server to go quite some time with no backups being generated and nobody realizing there was a problem!

One additional thought: Since I work in IT, I've simply set up a recurring reminder around time changes to check that things are working normally afterward, or in some cases to take action beforehand.  For an example of the latter, although in many cases getting an extra backup isn't such a bad thing, sometimes it can be.  In situations where you might only capture a Full every month or even every few months, getting two Fulls one after the other -- and therefore two runs of your retention policy -- could result in you losing your oldest Full and any of its child backups quite a bit sooner than you would have otherwise.  Or you might lose your oldest set to the disk space purge.  For cases where that risk exists, I'll set an alert for shortly before the time change.  At that time, I'll update the existing scheduled task to specify an end date before the time change, and then I'll create an additional scheduled task that has no end date but whose first occurrence is AFTER the time change.  Then before the next time change, I do it again.  In cases where an extra backup wouldn't create a problem, I'll just set an alert for after the time change to make sure that scheduled backups are still carrying on as normal if I'm not already set up to receive success email notifications

Great thoughts. Thanks again.

I AM set for successes as well as failures to be emailed. It only emailed ONE success for this task, however. I would not have noticed the additonal backup if I had not gone to my NAS to see the size and completion time. That's when I saw two within the hour and both started right on the hour.

It DOES give me one less retention set. Hopefully never a problem, but could be as you mention. You thought through this very well. Appreciated.

I'll assume this was it and check again on 1/1. If there are two then. I'll be back

~Bob

Hmm, you only got one email? Do the job logs of both runs mention sending an email? They should. Unless your email provider flagged one as spam or there was a transient issue that prevented one of them from being sent in the first place, I can’t imagine why you wouldn’t have gotten both emails. Reflect doesn’t skip emails based on how recently the previous backup occurred.

Interesting. Both logs show successful email sent. But I only recall seeing one. If I had seen two, I think it would have caught my attention. And did not go to SPAM. I have it white listed and check my SPAM daily.

Perhaps I am mis-remembering. But I don't think so.

We'll see what happens on 1/1

Must be the time change thing.  I encounter the same behavior with all Task Scheduler tasks set to run on Monthly schedules.  And it's always the same behavior: the 2nd instance of the task always runs 1 hour later, exactly, after the last run, run on its scheduled time.


Unfortunately, the only solution I've seen for this bug, since Microsoft won't address it, is it delete and recreate all tasks that do this.

I have run into a similar issue. Full backup scheduled for first Monday of the month at 23:00. In December, the full backup ran on the first Sunday at 23:00, then again on the first Monday at 23:00. Same thing happened in January (I didn't notice it until today). I checked task scheduler and the full task was only there once, scheduled for first Monday. Task history not enabled, so no help there.