Jakub, Thank you for the suggestion with nohup and I/O redirection. I'll have to see if that will work. Otherwise, we may have to keep the old behavior privately. Hong. On Tue, Aug 20, 2019 at 6:42 PM Jakub Jelen <jjelen@xxxxxxxxxx> wrote: > On Tue, 2019-08-20 at 16:00 +0900, Hong Cho wrote: > > Thank you for the reply. > > > > I changed the remote command to a generic example because in our > > case, we > > are using an internal binary that I didn't want to make public. I > > thought > > it wouldn't matter, but apparently, it does. I am sorry I can't share > > more > > about this tool. > > > > BTW, our environment is based on FreeBSD, not on Linux. In our case, > > because "isatty" is set to 0, it never closes. > > It never closes, because the `tail -f` never exists and never closes > the stdout and stderr that are connected to ssh channels. > > Yes, this is a difference from the previous openssh versions, but I > think this is a correct behavior. Previously, the ssh ran the command > in background and closed its IO. If you wish to run something in the > background, there are other ways to achieve (nohup and IO redirect), > which will work just fine also with OpenSSH 8.0: > > ssh user@xxxxxxxxxxx 'mohup tail -f /var/xxx > foo.out 2> foo.err < > /dev/null &' > > Regards, > -- > Jakub Jelen > Senior Software Engineer > Security Technologies > Red Hat, Inc. > > _______________________________________________ openssh-unix-dev mailing list openssh-unix-dev@xxxxxxxxxxx https://lists.mindrot.org/mailman/listinfo/openssh-unix-dev
