Gary Bunker wrote:
>
> Alexander-
>
> Yeah, pretty much EVERYBODY has the same problem. Netscape doesn't
> handle Java very well. Nearly everyone I know of that uses Linux turns
> off Java support in Netscape to avoid having to kill the rogue process
> every 30 minutes or so.
>
> On 10 May, Alexander Feigl wrote:
> > Hi!
> >
> > I have problems using Java pages with Netscape Communicator and MDK 7.0. (also
> > tried downloaded browser). NS crashes when loading a page with Java.
> >
> > Anybody have the same problem? Fixes?
> >
> > Alexander
U
linux-ld.so.2 is the library process which
netscape runs as a child of The freezing/crashing you
mention is the BEST terminating result--sometimes it just
thrashes the swap area and virtually locks you out--you might be
able to get a ctrl-alt-f2 recognized, but then logging in will
bomb on a password timeout because it will let you type VERY
slowly if at all. After 30-180 minutes it will generally pop up
with a login screen,,, What FUN!!!
I don't have a fix, but if you must try Java with netscape and
possibly open sites with Microsoft specific Java or poorly
written Java routines, then try this surgery
>---Begin on line below
#!/bin/sh
# routine to kill a zombie netscape without user intervention
# Save this as a text file and right click the icon, editing
# properties to make it executable, then drag it to the
# autostart folder.
# or else save it as a file and put ~/(filename) in rc.local
#--
#
# The method is to use ps and grep to pipe two lines to gawk
# which
# prints an output used by set to make the variable netshk equal
# to the process id of ld-linux-so.2
#(which runs netscape as a child)
# if the ps output lists netscape as a zombie process
#we KILL it
# Then, of course, we read the pid anyway and call top to
# measure cpu usage
#
# Currently it kills when it sees 98% or higher CPU
# this routine, which usually means it is freezing the screen
# or thrashing the swap file or both.
while [ 1 ]
do
set netshk = 0
set netshk = `ps ax | grep so.2 | gawk '/Z/{print $1}' `
if [ ${netshk} > 1 ]
then
kill -9 ${netshk}
fi
# End of zombie test
set netshk = `ps ax | grep .so.2 | gawk '/ld/{print $1}'`
# now have pid of netscape parent
set baddog = `top bn1 | grep .so.2 | gawk '/ld/{print $10}'`
# and the percentage of cpu time it is using
if [ ${baddog} > 98.0 ] then;
kill -9 ${netshk}
fi
# now we do not want to hog memory or cpu
# time ourselves
sleep 300
done
#
#-
>-end on line before this
This little Wild Netscape Hunter-Killer will end the annoyance of
freeze-ups from this cause at 5 minutes. There may be more
efficient ways to do this, but this one seems to work ... You
can echo messages to log files with each kill if you care to by
adding a line like
echo Netscape detected using ${baddog}% of CPU time and killed >>
or whatever your preference might be on a line after the kill and
before the fi.
That worked with 6.0 and 6.1
for 7.0 it is better to grep nets and gawk '/scape/...
for 7.1 we seem to be back to ld-linux.so.2
Naturally, no warranty. You Break it, you keep all the pieces.
And if use of this script transforms you into a poison dart frog,
then jump on Bill's shoulder.
Civileme
P. S. C&P is flaky again, so hope I managed to keep the typos
out of it.
