All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Windows 2016 Server - suddently I loose access to RDP on main srv and Win VM!
Hello, I have a private Windows Server 2016 server at Hetzner. I have had this server for 2-3 years now and have not had any issue with it.
But twice the last two weeks, I suddenly loose RDP access to the main server, but also loose RDP to my Windows VM (Hyper-V) on the same server.
But the strange thing is that my three Linux VM's on the same server, is up, and I have access to all of them.
All VM's have their own dedicated IP and each Hyper-V is only using the MAC on the NIC to get all the IP info it need to be online.
The first time it happened I just hit the hard reset button, and the server was up after the reboot. Today I got KVM access to see if I could see any issues.
When I used KVM I could see that the clock on the Windows logon screen was correct, and changed time correctly. But the CTRL - ALT - DELETE button and/or command from the KVM interface did not work.
I could move the mouse, but it would not access the logon window there I could type in my password.
The network icon showed that the server was online.
I had to hard reset the server, and then everything worked in the KVM after the reboot.
I have checked all windows logs, and since I today noticed the time when I lost access, I looked for all entries in all logs around that time, but there is no entries at all for all the time I did not have access. (total of one hour) It started to log again after the reboot.
So what can this be? I just loose RDP access to the main Windows Server installation and my Windows Server Hyper-V. But three Linux VM's running on the same Hyper-V is up and running, and there is no issues accessing them via SSH. And the websites and services is working fine.
All Windows logging is stopped for the duration of the "freeze". The clock is running fine, and I can move he mouse via KVM, but it do not accept any keyboard strokes via the KVM.
Any idea? This is way past my knowledge.
Comments
Try to disable Large send offload Ipv4 on your main server and Windows VM NIC and update your NIC driver.
Additional IP from Hetzner not always works well on Hyper-V, sometimes it will disconnect VMs for like 5 mins then it come back up (Event log showed that it wasn't able to retrieve IP from DHCP). Never had that kind of issue with locally routed subnet though.
I have now disabled it, so have to wait to see if this happen one more time.
But why will this issue stop all Windows logging and also freeze Windows so no commands from KVM (robot) is working?
And why do I just loose RDP access to the main server and the Windows Hyper-V server, and not the tree Linux VM's on the same server? (also using Hyper-V)
If Windows had freezed all together, the clock should have stopped on the lock screen, and if Hyper-V is freezing/stopping - all VM's should stop working, not just the Windows VM?
I have to say that I have very little network issues with this Hetzner. I'm using the Windows VM as a RDP desktop, that I always have up on my screen on my workplace. I never close the RDP connection, and the last year, I can't remember when the connection has been lost, beside when I patch Windows 10 and reboot the RDP desktop or when I reboot the main server because of Windows Update.
Did you only try once? I had issue with Hetzner KVM before as well, where I couldn't interact anything (so frustrated) and need to request them to replace it.
This look like the rdp related issue though, let see if disabling Large send offload will fix the problem or not.
Edit: you can also tick the option that allow unsecure connection via rdp on remote desktop configuration just for testing.
Depending on the kernel version you're using on your Linux VM's, they might not have the necessary paravirtualized syscalls that Windows has in the NT kernel.
I think starting in Windows Server 2012, Large Send Offload is disabled across all NICs by default.
I haven't checked Windows 2016 yet since I have never met problem on that version, but last time I checked on Windows 2012 R2 it was enabled by default :P
It was enabled as default on my Win2016 server also.
I tried one time, logged out from KVM, then logged in again, and still no response using any of the keyboard commands.
I then rebooted the server - still having the same KVM window up - and as soon as Windows started again, I could use the keyboard commands as normal in the KVM window.
But why do Windows stop logging anything in the systems log (system, application, security etc) when I can't login to the main server and Windows VM?
Edit: Now it's weird, not sure what happened tbh
Event Log on the main server don't show anything even a warning?
Nope, not a entry at all, the last entry was some minutes before, and was only a information under Security:
The WinHTTP Web Proxy Auto-Discovery Service service entered the run state.That was 15:22 my time, and I lost control 15:35 - 13 minutes after this last message.
I would see if there's a minidump file on \Windows\system32\ and go from there.
Minidump's are usually generated when a Windows NT based system Kernel Panics and can give you an insight on why the system blue screened, be it a hardware issue or driver issue
But in this case there was no blue screen - Windows was up and running, the clock was showing the correct time on the lock screen the whole hour.
And as told, I have 4 Hyper-V vm's on that server, and three of them was up and running the whole time and I could login via SSH on the three Linux servers.
And the webservices on the three Linux servers was also up and running. I have a owncloud server, and it worked fine, and I have two webservers with lots of sites on it, and they worked fine.
But the main server and the Windows VM did not let me in via RDP.
I could ping the server - so my Uptime monitor did not report the server as down.
It happened again today, and this time one of the three Linux servers also went offline, but the two other Linux servers was up and running and the website worked, and SSH access also worked.
But the main server and the Windows VM + one Linux VM was offline. I had to do a hard reset of the server to get access again.
And still no entries in the Windows log files.
I used your template from windowstemplate.com on a hetzner server, and lost remote desktop access too around the same date you made this post. It simply said the password was incorrect.
So I decided to reset the server to rescue mode.
Please let me know if you find a solution.
Then it's definitely something wrong here :S, Backup your data and do a fresh reinstall to see if it still happen maybe?
@AntKala if you not changed default password . Someone taken your server.
@myhken https://www.linkedin.com/pulse/hyper-v-virtual-machine-randomly-loses-network-under-windows-cheng
I did change the password of my admin account.
VMQ only affect Broadcom NIC though, Hetzner one is Realtek so I don't think that would be a problem
I have installed Teamviewer on the main server now - to see if I get access to that the next time. Network is up at least when this happen, since some of the VM's still is up.
Ok let us know how it goes.
It may be a RAM issue. Did just access my server and did see that the RAM usage was 98% - and that way to high. The server has 32 GB of RAM, and my 4 x VM's is using 22 GB RAM. The server is not running any other software using any high amount of RAM.
But still, the server was using around 9 GB of RAM, not used to Hyper-V.
Checked task manager, and no process was using more then 150 MB of RAM at the most.
So something is eating my RAM. I have never used more then around 80% of my RAM on that server.
If all the RAM is used up, can this cause this issue I describe?
I also did see that I had turned off the pageing file - I did now change it to min 1GB and max 16 GB.
Restart print spooler and see if ram usage goes down
If you run out of ram and have no paging file, process will get killed.
Got that issue before with 2016 (didn't exist in 2012). 99% sure it was the standby RAM that eaten up that much RAM.
Try this: https://wj32.org/wp/software/empty-standby-list/
Docs: https://www.guidingtech.com/57546/free-up-ram-windows-command-line/
I had a problem like that, but mine was an update.
Flushing standby ram only creates more work for the server. Standby is basicly cache. Leave it as is.
http://blogs.technet.com/b/askperf/archive/2010/08/13/introduction-to-the-new-sysinternals-tool-rammap.aspx