[comp.windows.x] X periodically dies on Sun 3.2

whm@arizona.edu (Bill Mitchell) (07/08/87)

I recently installed X V10R4 on our 3.2 Suns and 4.3+NFS VAXs and it seems
to work just fine for the most part.  However, during my usual mode of use,
running xinit on my Sun and rsh somevax xterm mysun:0, I get dropped back
out to where I started xinit from csh.  That is:


<login on Sun, get csh prompt #1>
xinit
<X starts up and I use it for a while>
<poof -- I'm looking at csh prompt #2>


There seems to be no pattern to when this happens.  Sometimes it'll last all
night, sometimes less an hour or so.  It's gotten me both after long periods
of inactivity and between keystrokes.

I've tried checking $status after xinit exits, but it's 0 and there's no
core dump that I can find.  Often, when I try to start another xinit after
getting dumped out, xinit will say "No such display server unix:0".  I found
that if just persist, xinit eventually works.

I like X a lot, but it's really irritating to have a bunch of windows up,
some su's going, an Emacs or two, and then have all that blotted out.

Any ideas or suggestions will be greatly appreciated.

					Bill Mitchell
					whm@arizona.edu
					{allegra,ihnp4,noao}!arizona!whm

mende@aramis.rutgers.edu (Bob Mende) (07/08/87)

>  [ XV10R4 dies sporadically]	

The only time that thise type of thing happens to me is under uwm (and
menuwm) when I have some window like emacs or xterm popping up and I
am holding down one of the mouse buttons to do somthing with the
menus.  Once and a while (cant determine exactly why) this will kill
uwm and the X server and return me to where I invoked xinit (which I
exec most of the time!)

				Bob
-- 
     mende@rutgers.edu    {...}!rutgers!mende    mende@zodiac.bitnet

bob%tut.cis.ohio-state.edu@osu-eddie.UUCP (07/09/87)

In article <1801@megaron.arizona.edu> whm@arizona.edu (Bill Mitchell) writes:
>I recently installed X V10R4 on our 3.2 Suns ...  running xinit on my
>Sun and rsh somevax xterm mysun:0, I get dropped back out to where I
>started xinit from csh.  That is:
>
><login on Sun, get csh prompt #1>
>xinit
><X starts up and I use it for a while>
><poof -- I'm looking at csh prompt #2>
>
>There seems to be no pattern to when this happens.  Sometimes it'll
>last all night, sometimes less an hour or so.  It's gotten me both
>after long periods of inactivity and between keystrokes.

	Two of our users have complained of the same phenomenon.  They
think it may be after long periods of typing into a remote client
without any other activity on the local server - particularly, no
mouse motion.  It seems that when they jostle the mouse occasionally
the symptom isn't noticed.

	It's particularly frustrating to me because I haven't seen it
myself (I mouse a lot), and they can't really provide me with
reproduce-by scripts.  I suggested the gross hack of using
XWarpMouse() to (virtually) move the mouse in little circles, one step
every fifteen seconds or so, and observing the effect.  No reports
back as to whether this changed anything, but you might try it.
-=-
 Bob Sutterfield, Department of Computer and Information Science
 The Ohio State University; 2036 Neil Ave. Columbus OH USA 43210-1277
 bob@ohio-state.{arpa,csnet} or ...!cbosgd!osu-eddie!bob
 soon: bob@aargh.cis.ohio-state.edu

bob%tut.cis.ohio-state.edu@osu-eddie.UUCP (Bob Sutterfield) (07/14/87)

In article <3777@osu-eddie.UUCP> bob@ohio-state.ARPA (Bob Sutterfield) writes:
>In article <1801@megaron.arizona.edu> whm@arizona.edu (Bill Mitchell) writes:
>>
>>I recently installed X V10R4 on our 3.2 Suns ...  running xinit on
>>my Sun and rsh somevax xterm mysun:0, I get dropped back out to
>>where I started xinit from csh.  That is:
>>...
>>There seems to be no pattern to when this happens.  Sometimes it'll
>>last all night, sometimes less an hour or so.  It's gotten me both
>>after long periods of inactivity and between keystrokes.
>
>	Two of our users have complained of the same phenomenon.  They
>think it may be after long periods of typing into a remote client
>without any other activity on the local server - particularly, no
>mouse motion.  It seems that when they jostle the mouse occasionally
>the symptom isn't noticed.

Well, just to call off the bloodhounds who may or may not be looking
for this one (like we were) in the wrong place (like we were):

A couple of local folks have developed extensions to csh to support
EMACS-style command line editing and TENEX-style completion.  Among
the other extensions was a shell variable called `autologout'.  When
autologout is not explicitly unset, or when it is set to some non-zero
value, the shell will kill itself after a period of inactivity.

And sure enough, the console window, closed up and tucked away up in
the corner, often has long periods of inactivity.  And when that shell
kills itself, being the default initial client, it takes the whole
server with it, in a spectacular death rattle of `XIO: socket not
connected' messages.

This may or may not be the problem at other sites, depending upon
whether you have this sort of an extension in your csh.  We suspect
that it was the problem here.
-=-
 Bob Sutterfield, Department of Computer and Information Science
 The Ohio State University; 2036 Neil Ave. Columbus OH USA 43210-1277
 bob@ohio-state.{arpa,csnet} or ...!cbosgd!osu-eddie!bob
 soon: bob@aargh.cis.ohio-state.edu