The STDERR messages seem to be more alarmist than necessary, and give me a false feeling that xpra itself may have failed, or that some critical feature has failed, when such does not seem to be the case. The following sample paste of an xpra STDERR illustrates with two examples:
1511 $2018-01-03 00:02:08,828 server failure: disconnected before the session could be established 2018-01-03 00:02:08,830 server requested disconnect: server error (error accepting new connection) 2018-01-03 00:02:08,846 server failure: disconnected before the session could be established 2018-01-03 00:02:08,847 server requested disconnect: server error (error accepting new connection) Warning: cannot use the system proxy for 'start' subcommand, FAILURE
1] The initial messages indicate that the server experienced a failure, but the server seems fine.
2] The final message issues a warning, followed by a cryptic (and alarming) "FAILURE".
This choice of words does xpra a dis-service, in that it makes me (and probably other users) unsure and uncomfortable with the product's stability and functionality.
My suggestion is to reserve the word 'failure' for situations that render the software unusable. Maybe otherwise use the word "error". Alternatively, a term such as "non-critical failure" might be appropriate.
This warning message should only be shown when the system service is not available, this should not happen with the packages we provide. The "FAILURE" message is a generic lookup of the sub-command's exit code, I'll try to special case it to make it more user friendly.
I cannot reproduce the warning message you are seeing:
Warning: cannot use the system proxy for 'start' subcommand, failed to connect to '/run/xpra/system': [Errno 111] Connection refused
With more details in debug mode. This is fine.
/tmp/hello):
Warning: cannot use the system proxy for 'start' subcommand, failed to connect to '/tmp/hello': [Errno 2] No such file or directory
And this also looks fine to me.
So, there must be something different going on with your system. r17839 changes the wording to "unknown general failure" because that's all I know about your specific failure state.
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/1733