For some reason, after LOTS of idle chatter between my F21 SSH client and the server (I am assuming these are SSH application keep-alives), my F21 SSH client sends a TCP keep-alive. This is after some long time of the client NOT sending an SSH idle packet.
As soon as the server receives this TCP keep-alive, it sends a TCP RST and that 'breaks the pipe'.
So what in the client determines the rate of SSH keep-alives and for how long inactive?
Then with the app level inactive, why the TCP keep-alive?
Or do I take this question to the OpenSSH list?
Based on the symptoms you describe in this email and your earlier ones, I would most strongly suspect that one of the systems you're using has an IP address conflict. Use arping on both the client and server to verify that they are the only device on their respective broadcast domains to use their address.
On 03/18/2015 05:30 AM, Robert Moskowitz wrote:
For some reason, after LOTS of idle chatter between my F21 SSH client and the server (I am assuming these are SSH application keep-alives), my F21 SSH client sends a TCP keep-alive. This is after some long time of the client NOT sending an SSH idle packet.
As soon as the server receives this TCP keep-alive, it sends a TCP RST and that 'breaks the pipe'.
So what in the client determines the rate of SSH keep-alives and for how long inactive?
Look at ServerAliveCountMax and ServerAliveInterval in /etc/ssh/ssh_config and ~/.ssh/config, however as Gordon has mentioned, it smells like an IP conflict and one way to sort it is "arping" the remote host.
<snip> ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer, AllDigital ricks@alldigital.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - - "How does that damned three seashell thing work?" - - -- Sylvester Stallone, "Demolition Man" - ----------------------------------------------------------------------
On 03/18/2015 01:22 PM, Rick Stevens wrote:
On 03/18/2015 05:30 AM, Robert Moskowitz wrote:
For some reason, after LOTS of idle chatter between my F21 SSH client and the server (I am assuming these are SSH application keep-alives), my F21 SSH client sends a TCP keep-alive. This is after some long time of the client NOT sending an SSH idle packet.
As soon as the server receives this TCP keep-alive, it sends a TCP RST and that 'breaks the pipe'.
So what in the client determines the rate of SSH keep-alives and for how long inactive?
Look at ServerAliveCountMax and ServerAliveInterval in /etc/ssh/ssh_config and ~/.ssh/config, however as Gordon has mentioned, it smells like an IP conflict and one way to sort it is "arping" the remote host.
There is no ~/.ssh/config and neither of these variables are in the /etc/ssh/ssh_config, so they are whatever the defaults are.
These servers are on my internal subnets. I know exactly what boxes are on which subnets and what their addresses are. But I checked anyway, and there are no IP conflicts.
thank you for your help
-
Gordon Messmer -
Rick Stevens -
Robert Moskowitz