Recently we upgraded from server 5.7.4 to 6.1.0. The next day (12/5/19) at approximately 3:50pm server time the jobs.xml file stopped getting updated with the appropriate next/prior run times of our active jobs. This led us to believe that the server was off or not running our scheduled jobs. But when we opened the client we could see all the correct times and the jobs were running as scheduled. There is some odd logging at 3:50 but no explicit error. We first tried stopping and starting the server to fix this unsuccessfully, so we restarted the service manually and after that jobs.xml began updating normally.

Thanks for your time.

Please send the log_serverDATE.txt from the log folder to support@visualcron.com. Please also check the event log for any unusual events during that time.

---

I've attached the log. And there was nothing suspicious in the event logs at that time. Also I think I confused myself and the file actually stopped being updated at 3:50AM rather than PM. Which makes me think what I thought was odd in the log at 3:50pm was unrelated. My mistake, sorry.

Does not seem to be the right file. I see no errors or stop in log. I do not see the restart of service either.

---

Sorry, the restart wasn't until today. Here is all the logs from the interim. Sorry for any confusion.

Okay, so it happened again today. Last update to jobs.xml was at 2:40am server time. Attaching the logs.

When looking at the file you can see that it updates around the clock. It does not stop updating. What problem do you experience exactly?

---

We experience no actual problems with scheduled job being properly run. Everything from the client side seems to be updating and running normally and the jobs are definitely being run. But these updates are not being reflected in the jobs.xml file. Specifically, the DateNextExecution and DateLastExecution elements for every job and task are not changing they cease to be updated in jobs.xml at 2:40am. But in the client they look fine. If I make a manual change to a job name or description it is reflect in jobs.xml but those elements remain unchanged. These elements are updated consistently on our other servers running 6.1.0 so I assumed this is not expected behavior. We discovered this because we use the jobs.xml file to monitor VC and confirm the jobs are being run. When we found that DateLastExecution and DateNextExecution weren't changing for any of the jobs it was concerning. In the log there is a blip at 2:40am where the logging is abnormal. This leads me to believe that something went wrong that affected the updating of the jobs.xml file and possibly other .xml settings files that I didn't notice.

Ok,

1. what happens if you, after last run, wait a minute (until you see Jobs saved). Then restart the service, close client and reconnect. It the last run correct then?

---

I tried it several times and the last execution didn't seem to change, but when I would look at the job log (the right-click on job one) there seemed to be a gap between 3:50pm and 4:07pm (I restarted the service at 4:07pm) where even though jobs definitely ran at that time those runs' event time, duration and exit codes do not show up within that log. I'll attach the updated log.

4:07 was when I first restarted the service. I tried restart again twice between 4:07 and 4:14, and I saw no effects on the last run times or the job specific logging caused by those restarts.

As an update,

We tried uninstalling and re-installing multiple times and restarting the machine multiple times, but after a day or two jobs.xml again would stop being updated. On a hunch I inactivated several inessential jobs with five minute interval triggers that I kept seeing being run at the times the file stopped being updated. Since that change it's been almost two weeks without seeing the error. I don't really know why it worked, but it seems like it has.

As an update,

We tried uninstalling and re-installing multiple times and restarting the machine multiple times, but after a day or two jobs.xml again would stop being updated. On a hunch I inactivated several inessential jobs with five minute interval triggers that I kept seeing being run at the times the file stopped being updated. Since that change it's been almost two weeks without seeing the error. I don't really know why it worked, but it seems like it has.

Thank you for the update. When I look at the log file again I see constant updates of "Jobs saved". We do log if any failure should occur during the actual saving but there is no trace here.

One thing that might be interesting for you to monitor is how we save Jobs. What we do is the following, in order;

1. place a temporary lock on the Save method
2. serialize (convert current list of Jobs to xml format) into a temp file called jobs.tmp.xml
3. delete the current "jobs.xml"
4. rename the jobs.tmp.xml to jobs.xml
5. release lock

The same object that is used to create the file is also used when you, for example, connect with a new Client. So, it would be very strange if the file is not updated while the newly connected Client can see the updates.

We do seldom update the Job instantly. Normally a minute after changes has been made.

---