So I've had this issue for a while now, but never really bothered to research/question it until now. Basically what happens is random servers (all running w2k3) will just stop responding to RDP requests after a reboot even on different hosts. Now this has happened on numerous datacenters that I have implemented - ESX 3.0 right through to ESX 3.5 update 2. The only way to fix is to reboot right after - as such I usually get the guys to check the RDP conn when patches are done etc etc, but sometimes they forget and then we need to log a change to get the server rebooted.
The VM's will respond to ICMP, network shares etc. its just the RDP thats broken!
I've done some research, but no one seems to have an answer to this....this issue only comes up on virtual enviroments. I havent seen this happen on physical servers in a datacenter of 100+ physical boxes ranging from Win 2000 - 2003!
Has anyone out there experienced the same issue?
I just now got aroudn to researching this issue too. We have this at many of our clients. Reboot the server and RDP is dead. No event log errors etc. Reboot it again and 99% of the time RDP is back in business.
THe common denominator seems to be Windows Server running on slow disk access boxes. We have a physical server at a client that is just beat down and it experiences this behavior about every other reboot. Obviously the overahed that the other VM's create is probably causing the issue when it is a VM.
Be good to know if setting a time out value somewhere would fix this issue. I do believe that the TS service is not starting properly due to some strain on the disk i/o causing it to time out.
I am also seeing this issue every once in a while. Would be good if we can figure out the root cause of this.
After reviewing the above linked KB article ( http://support.microsoft.com/?kbid=930045 ) I think I might be hitting it. Perhaps something about running in a VM tickles the bug, but it does seem to be a Windows issue.
From what I've seen this only happens on machines with a lot of disk i/o. (i.e. Soft mirror of 2 sata or IDE drives. Sometimes SCSI.)
I have this on a lot of low-end SBS boxes.
Only fix I could think of would be to increase the timeout value on the Terminal Service service. Anyone got any leads on this?
Jason Redwine, MCSE, CCNA, CCA
LAN RM Technologies
250 East 5th Street
Cincinnati, OH 45202
Office & Fax (513) 984-3900
Cell (513) 616-6215