In short, getting this:
sh: 1: /run/user/1000/xpra/run-xpra: Permission denied
on Ubuntu 12.04, where /run is mounted by default as:
tmpfs on /run type tmpfs (rw,'''noexec''',nosuid,size=10%,mode=0755)
I can of course modify this in fstab, but filing ticket as this is the default on Ubuntu. Either run-xpra should remain in $HOME, or some test should be done to check whether $XDG_RUNTIME_DIR is mounted noexec.
(Note that /run is mounted in the initramfs, so a hack in fstab to change this actually requires an entry containing "remount" in its options, since /run is not listed in fstab normally at all.)
Original feature ticket: #1129
I've re-tested 1.0 on all those distributions to verify:
[/var]/run/user
does not exist
So the only problem that I can see is with 14.04, not 12.04, and it isn't a huge problem since we try to execute ~/.xpra/run-xpra
first.
What is the problem you're seeing?
Sorry, it was a typo; I am in fact on 14.04 (trusty).
The problem is occurring when there is neither a ~/.xpra or $XDG_RUNTIME_DIR/xpra directory yet existing on the destination host. The autogenerated run-xpra script gets created in $XDG_RUNTIME_DIR in this case.
Looks like changing 'bind' from 'auto' to '~/.xpra/' makes the script appear in ~/.xpra as with previous versions. I suppose the least overhead solution to this issue would be a documentation note, or comment in the shipped 50_server_network.conf, that Ubuntu 14.04 systems should use that setting.
don't use XDG_RUNTIME_DIR for run-xpra script on Ubuntu 14.04
Forgive me if I am being thick but I still fail to reproduce any kind of showstopper problem on 14.04:
rm -fr ~/.xpra ${XDG_RUNTIME_DIR}/xpra
Resets things for testing, then I can use any xpra command I like and when the "run-xpra" scripts get created (xpra start, xpra shadow, etc), they are created in both locations.
If I use a command that relies on those scripts (ie: xpra start ssh:HOST --start=xterm
), then it finds the one in "~/.xpra" first and the command runs without problems.
What exact commands are you using to get a failure?
Does the patch attached above fix things?
Not heard back, closing.
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/1394