For some time now we have been having regular occurances of a connectivity issue with our SMP server, currently running 7.6 HF1 but this occured back in various releases of 7.5 also. The symptoms are:
1. Console would open but any pop-ups using Java would not load i.e. driver management and pre-defined computer configuration
2. Could not RDP on to server, or when it did work only with a local account
3. Scheduling of jobs would sometimes work but in most cases would not
4. Telling the VM to restart would often cause it to completely hang
Forcing the server off and back on again was often the only way to "fix" this. But the other week I noticed something, and that was that the times of the day it occured it was during the period that it was attempting to push SMA out to the network, configured for every 2 hours.
Once I noticed this and rescheduled it to to it only once per day, in the afternoon, I noticed that these periods of inactivity were less frequent and in fact occured at the scheduled time of the SMA push.
The question then is why is the server becoming un-responsive in this way? Is there a way of limiting the number of machines that it attempts to push to at any one time. My guess is that the majority of the machines that it is pushing to, synced from AD are either old machines that have not been turned on in some time, or machines that have failed, or indeed de-commissioned machines that we have not removed from AD yet. These machines will not respond and therefore leave a load of half open connections.
In any case I wouldn't expect the server to hang like this and yet it does. The VM is configured with the VMXNET 3 adapter and has a 10GB connection to the network, as do each of the ESX hosts that we have, so I very much doubt it is saturating this link.
Any insight into this would be most helpful.
Howard