Vlad_Navazhylau_6186Reading the "turnover" description, I don't think that's it, because it sounds like an issue where it simply overflows with the RunIDs (possibly reaching the limit for the database field) and restarts them at "1".
Based on what Josef said, the explaination is likely in different machine times.
I can only weigh in for Linux (Windows will not be all that much different) but keeping time among agents and the server is an interesting issue. Even if you have an NTP server, each machine has a hardware clock, possibly a different BIOS clock, and a running clock that is kept by the kernel. The last one is the only remotely useful one (the HW clocks are horribly unstable with time keeping), so if your Automic agents starts before the machine has synced with the NTP server after boot, that'd expain a time difference. With UNIX, you even have two competing ntp daemons (OpenNTPd and Berkley ntpd) that work different, and those correct the clock in different ways, e.g. small clock differences are "skewed" by making the local clock run slightly slower or faster for a while, while bigger differences are compensated in other ways.
Your 18 hour difference, for instance, might have been (just a theory) a server with a dying CMOS battery that fails at keeping time when it's shut off, gets booted, and executes a job before NTP kicks in. The OS has no choice but to use a legacy hardware clock before it gets a proper time source.
Take all that together, and that probably explains the time differences. The missing puzzle piece may have been provided by
Josef_Scharl_103 in that the timestamps are taken off of different machines.
But on a closing note, I don't think it's the greatest design decision by Automic either to use time from different machines in the DB statistics. I'd consider it much cleaner design for the Agent to tell the server that it has begun executing the job, and the server then logging the time from it's own time source *)
Cheers,
Carsten
*) edit: or even log both times. That'd be ideal for debugging, but possibly promote DB bloat.