Connectivity issue to Google Cloud VM via RDP
446 views
Skip to first unread message
Костя Хомко
Dec 8, 2017, 11:33:03 AM12/8/17
to gce-discussion
Hello,
We are facing with connectivity issue to Google Cloud VM each day.
Some of our servers are not reachable and we can not connect to them from 1-st and sometimes even from 10 attempt whereas these servers are up and running.
This issue is reproducible for all our servers which are located rather close for us and even for those which are even far from us.
All that we can see when it happens - only black screen and attempt to reconnect.
Any ideas?
We are facing with connectivity issue to Google Cloud VM each day.
Some of our servers are not reachable and we can not connect to them from 1-st and sometimes even from 10 attempt whereas these servers are up and running.
This issue is reproducible for all our servers which are located rather close for us and even for those which are even far from us.
All that we can see when it happens - only black screen and attempt to reconnect.
Any ideas?
Костя Хомко
Dec 8, 2017, 12:34:47 PM12/8/17
to gce-discussion
Short update. We currently using Windows 2008 R2 OS on all of our servers.
Carlos (Cloud Platform Support)
Dec 8, 2017, 4:09:30 PM12/8/17
to gce-discussion
Here are some ideas that might help you to isolate the issue:
a) Try different networks and RDP clients
b) See if there are error messages in the Serial Console of the servers.
c) Keep a log to try to determine patterns on times, servers and locations when this happens.
d) Use ICMP and traceroute to gain visibility over the network
e) You could spin a linux VM in your project, install telnet and test connectivity to the RDP port (i.e. telnet x.y.z.w 3389) when you are experiencing the issue. That would also help you discard a network problem.
f) Check utilization resources in your servers when this happens Cloud Console-> Server (i.e. CPU, memory).
g) Follow this Microsoft troubleshooting guide and check this discussion.
h) Turn on some RDP auditing in your servers to add visibility
Since there are many components involved and your problem is intermittent it might take some effort to isolate the problem.
a) Try different networks and RDP clients
b) See if there are error messages in the Serial Console of the servers.
c) Keep a log to try to determine patterns on times, servers and locations when this happens.
d) Use ICMP and traceroute to gain visibility over the network
e) You could spin a linux VM in your project, install telnet and test connectivity to the RDP port (i.e. telnet x.y.z.w 3389) when you are experiencing the issue. That would also help you discard a network problem.
f) Check utilization resources in your servers when this happens Cloud Console-> Server (i.e. CPU, memory).
g) Follow this Microsoft troubleshooting guide and check this discussion.
h) Turn on some RDP auditing in your servers to add visibility
Peter Hornyack
Dec 11, 2017, 5:26:41 PM12/11/17
to gce-discussion
Another possible problem is that the MTU of your Windows instance has been set to too high a value. The Google Compute Engine network supports an MTU of 1460 or less; I've seen RDP connections fail when the MTU is accidentally set to 1500.
To check and lower the MTU on a running instance, follow these instructions to access the Windows SAC on serial port 2, start and connect to a CMD prompt from the SAC, then run 'netsh interface ipv4 show subinterfaces' and 'netsh interface ipv4 set subinterface "<interface name>" mtu=1460 store=persistent'.
Костя Хомко
Jan 4, 2018, 5:19:45 AM1/4/18
to gce-discussion
Thanks. Changing MTU from 1500 to 1200 solved an issue!
Reply all
Reply to author
Forward
0 new messages