I have a virtual environment for four servers. Every so often, three of them loose 30 minutes of time for no good reason. I have tried all sorts of things to stop this but it keeps happening.
I am at a bit of a loss as to firstly why virtual servers would loose time at all and secondly, why it doesnt affect all servers!
What virtualisation environment and do you have host time sync enabled. If you do it can end up competing between the network NTP time and the dominant host sync time. These two conflicting can easily cause issues especially if your AD servers are virtualised.
You can keep the clients' time updated by:
1. The tools installed (e.g. VMTools in the case of VMWare). This periodically updates the host's clock to that of the client's clock.
2. A time server. This is by far the best way of doing it.
This article explains a bit of the background and why you experience drift.
In a nutshell, just configure your client machines to use a time server.
When my time was drifting, with a virtual DC as the primary time server, it was because the physical boxes the virtual stack ran on were getting their time from the virtual DC. Setting them to get time data from the NTP pool, same as the DC, sorted it.
its a VMWare setup - i think ive sussed it though. Ive changed the NTP server the ESXI box was pointing to and just used uk.pool.ntp.org and it seems to be better!
Make sure to switch off VMware client time sync to otherwise they will still be fighting each other and this will cause drift over time.
Originally Posted by mtillbrook