Very strange - system time changes with 6.5

[cliffm]

Have had some bizarre issues since upgrade to 6.5
6.1 on xp 32 bit working just fine. Had surebackup issues but these were more or less solved, still the subject of a case with support.
I have replicas set to run every 2 hours. I set the replicas to run on the hour. After upgrade to 6.5 the replicas ran 2-4 times in succession each time the schedule came up. This has been the case for the last week. This has been the subject of a new job logged with support but I don't have the answer yet. So today I changed the schedule for the replicas to run at 2 minutes past the hour. This worked, now the replicas only run once. BUT each time they run 10 hours gets added to the Veeam server system time. So the jobs run for approx 10 minutes but the jobs end 10 hours and 10 mins after they start unless I change the time back while the job is running! I have tried VMware tools time syncing to host turned on and off. Same problem. The support job # is Case # 00161031. I have been waiting for a webex all afternoon but have not had a response from support yet.

Anyone have any ideas?
Thanks

[tsightler]

Is one of the replicas you Domain Controller? Have you checked the time on the ESXi host? Even with VMware tools time sync disabled, some operations (snapshots for example) still force time sync with the host so it is important that the ESXi host time be set correctly. In at least one case I knew of a customer that was replicating/backing up their domain controller. When the snapshot was taken, it was changing the time on the domain controller, and then other systems in his environment would sync to the "wrong" time on the DC. In the end, you need to make sure that the ESXi hosts have to proper time settings in all cases.

This thread has some details:

http://communities.vmware.com/thread/237437

Note the second post which explains that snapshots are one of the events that are not affected by disabling VMware tools timesync. There is an advanced parameter that can added to the VMX file to disable this behavior, but the best solution would be to verify the time on the ESXi host as I've never seen this issue when that wasn't the problem. Although, there's of course a first time for everything I guess.

[cliffm]

Yes the replica is the domain controller. The time on all hosts is correct, they sync with NTP. The time on the domain controller is correct. All other domain members are displaying the correct time. The Veeam server is a domain member as well, this has never happened before and has only occurred today after changing the replication schedule.

I found the issue. We have a router that for the first time in the history of history is not updating correctly via NTP. It is 10 hours fast despite pointing to local NTP servers. Somehow changing the veeam schedule has set the veeam server to get its time from this router. At least I think that is what has happened. No other machines are affected.

So problem worked and solved. Thanks.