I have been a bit suspicious about SELinux myself, the system is running in
enforcing mode. So I temporarily changed it to permissive mode, but that had no
effect. So this is not an SELinux problem.
The server is a fairly recent installation and I've deliberately tried to keep
it as close as possible to plain RHEL8, with a minimum amount of configuration
changes. The only real change made to it is VirtualGL, a package for
server-side 3D rendering that integrates with Thinlinc. But this does not seem
like a graphics problem, does it?
I'll try the new appImage version once it's available.
Best
Anders
________________________________
Från: Thomas Friedrichsmeier
Skickat: Lördag, 30 juli 2022 17:50
Till: Anders Fridberger
Kopia: rkward-devel@kde.org
Ämne: Re: rkward-devel Digest, Vol 86, Issue 12
On Sat, 30 Jul 2022 12:27:05 +0000
Anders Fridberger <anders.fridber...@liu.se> wrote:
> The sockets are created OK, with the same permissions as yours
> (srwxrwxr-x. 1 anders anders 0 Jul 30 14:18 rkwardOT1p6oMh). I
> can connect to the rkwardXXXXX socket as well as the rkdXXX one (no
> errors), but the rkdXXX socket is removed fairly soon after it
> appears (presumably when the backend crashes).
That's an interesting observation, although I am not sure that it would
get us anywhere. Both sockets are actually created by the frontend, and
waiting for the backend to connect. Each would disappear (close), as
soon as the frontend receives a successful connection. Thus, if all
went well, the rkwardXX socket would be expected to disappear within a
fraction of a second, while the rkdXX socket should stay
around until either something is plotted, or the frontend
exits.
Despite lack of evidence, I still suspect a permission problem. Is
SELinux activated? I have no first hand experience with managing that,
but from my understanding, in this case
ausearch ausearch -m AVC,USER_AVC,SELINUX_ERR,USER_SELINUX_ERR
-ts recent
after attempting to start rkward should list a denial, if this is the
problem.
Are you aware of any other configuration details that might relate?
Beyond this, I have added a few more lines of debug statements, hoping
to understand better, what's going wrong. Unfortunately, the daily
AppImage build has already started, the addition info would be expected
in build #44 in about 24 hours. (Alternatively, you could build from
git master branch).
Thanks a lot for your continued help in tracking this down!
Thomas
Regards
Thomas
