[slurm-users] slurm-17.11.5 usage of X11

[John Hanks]

Hi Philippe,

The built-in --x11 support still has some rough edges to it. Due to a
timeout issue with getting xauth set up, I've been patching SLURM with
these minor changes:

--- src/common/run_command.c_orig 2018-02-06 15:03:10.000000000 -0800
+++ src/common/run_command.c 2018-02-06 15:26:23.525128464 -0800
@@ -198,7 +198,7 @@
        __func__, script_type, max_wait);
  break;
  }
- new_wait = MIN(new_wait, MAX_POLL_WAIT);
+ new_wait = MIN(new_wait, 10000);
  }
  i = poll(&fds, 1, new_wait);
  if (i == 0) {


--- src/common/x11_util.c.orig 2018-01-19 12:51:10.400391579 -0800
+++ src/common/x11_util.c 2018-01-23 08:00:42.080557637 -0800
@@ -146,7 +146,7 @@
  xauth_argv[1] = xstrdup("list");
  xauth_argv[2] = xstrdup(getenv("DISPLAY"));

- result = run_command("xauth", XAUTH_PATH, xauth_argv, 100, &status);
+ result = run_command("xauth", XAUTH_PATH, xauth_argv, 10000, &status);

  debug2("%s: result from xauth: %s", __func__, result);


The longer wait makes it work almost everywhere on our system. However, I
still have a small number of users for whom it fails and we still haven't
figured out why.

Best,

jbh

On Wed, Apr 11, 2018 at 12:56 AM, Philippe Grevet <
philippe.grevet at ips2.universite-paris-saclay.fr> wrote:

> Hello,
>
> We have some problem to use native X11 of  the new slurm version (17.11.5)
> compiled from source on Debain stretch.
> Before we use the plugin spank with no problem on an older version of
> slurm.
>
> I have first this error
> "error: X11 forwarding not built in, cannot enable."
>
> After reading different mails,i add
>
> in slurm.conf
>
> PrologFLags=X11
>
> and in /etc/ssh/sshd_config (on  head login server of the cluster)
>
> X11UseLocalhost=no
>
>
> The error change and become
>
> X11 connection rejected because of wrong authentication.
>
>
> X11Forwarding is working because if i first open a interactive session,
> and then connect whith ssh -X on the node, the x-application is working
> fine and displayed on my client computer.
>
> Then I look at .Xauthority file
> ( first i removed it and recreate a new)
>
> #ssh -X me at cluster
>
> #echo $DISPLAY
> cluster:11.0
>
> #xauth list
> cluster.ips2.u-psud.fr:11  MIT-MAGIC-COOKIE-1
> 2809a05f53f693e416682f4befa76a39
> #
> $ srun    --pty bash --init-file /etc/profile
>
> me at node4:~$ echo $DISPLAY
> cluster:11.0
> #
> #$ xauth list
> node4.ips2.u-psud.fr:11  MIT-MAGIC-COOKIE-1
> 2809a05f53f693e416682f4befa76a39
>
> ###the hostname is now the hostname of the node
>
> me at node4:~$ xeyes
> X11 connection rejected because of wrong authentication.
> Error: Can't open display: cluster:11.0
>
> ###But if i add :
> #
> me at node4:~$ xauth add cluster.ips2.u-psud.fr:11  MIT-MAGIC-COOKIE-1
> 2809a05f53f693e416682f4befa76a39
>
> #I now can run x-application
>
> Have you an idea of the problem ?
>
> (User login/password authentification of the cluster are done again an
> Active Directory annuaire with sssd )
>
>
>
>
>
>
>
> *Philippe GREVET*
> philippe.grevet at ips2.universite-paris-saclay.fr
> Tel   *0169153393*        Bureau - 241  Bâtiment -630
> http://www.ips2.u-psud.fr/
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.schedmd.com/pipermail/slurm-users/attachments/20180413/91d4ee36/attachment-0002.html>