On 8/20/17 4:37 PM, C.v.St. wrote: > BUT, if you loose the 'mosh client process', which is the only > instance to connect to your running session, this 'session' is > definitely lost and dead, even if the programs in it stay > alive for a while. Also, if your program's output is more > than the buffers between program and mosh-server hold, the > program(job) will be stopped waiting for mosh.
That's not quite how we think of it, or expect it to work. If your mosh-client dies (by your machine crashing, or running out of battery), the mosh-server continues to run and the session and programs in it should be unaffected-- you just can't reach the terminal session any more. I think of that as "alive but lost". Also, mosh-server doesn't actually "buffer" output from your program-- instead it updates a virtual terminal display, and your program shouldn't ever have to stop and wait because the client is slow or unreachable. We've checked this on Linux and MacOS, and it behaves this way. If your remote program really does stop if your mosh-client is unreachable/dead, would you please open an issue on GitHub? (I do agree that long-running jobs should be run in 'screen' or 'tmux'-- those programs offer some protection against accidental termination of the session, even when using Mosh.) regards, --jh _______________________________________________ mosh-users mailing list mosh-users@mit.edu http://mailman.mit.edu/mailman/listinfo/mosh-users