[FreeNX-kNX] nxagent sessions left headless (was: Load Balancing leaving unconnectable sessions)

[Matthew Richardson]

I've looked at some more of these stuck sessions, and I think in some
cases it was caused by having the connection timeout set too low.

However, I'm seeing another problem occurring on several 'slaves', where
the nxagent session (and all children) are running headless.  The
session log on the slave shows, e.g:

...snip... (normal startup stuff)
Session: Session started at 'Mon Aug  3 22:49:07 2009'.
Warning: No data received from remote proxy within 30 seconds.
Error: No data received from remote proxy within 120 seconds.
Session: Display failure detected at 'Tue Aug  4 01:16:57 2009'.
Session: Suspending session at 'Tue Aug  4 01:16:57 2009'.
Session: Session suspended at 'Tue Aug  4 01:16:57 2009'.

The pool master has this session's file in nxserver/db/closed/ dated
'Tue Aug 4 07:24am'.  There are no processes associated with this
session left on the master, and any new user sessions will be spawned
elsewhere in the pool.  Over time, this will obviously chew up system
resources, and might cause clashes if the master happens to reallocate a
new session to this slave (we run 1 session per user to help keep track
of these kind of problems).

I've seen lots of references to this warning message from 2006/07, but
with no obvious solutions.  any ideas?

Thanks

Matthew

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 260 bytes
Desc: OpenPGP digital signature
URL: <http://mail.kde.org/pipermail/freenx-knx/attachments/20090817/be4f6104/attachment.sig>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: not available
URL: <http://mail.kde.org/pipermail/freenx-knx/attachments/20090817/be4f6104/attachment.ksh>