sma

When I was running R 1.6.1 (under Red Hat 8.0), R would abruptly 
terminate from time to time. I would type in a line of code and hit 
return, and the cursor would not move from the end of the line. Then the 
hard drive would start churning. This goes on for at least 10 minutes. 
Finally R terminates (finally freeing up the keyboard and display). The 
only word printed is "Killed" immediately at the end of the last line of 
code. I upgraded to 1.6.2 and the same thing continues to happen. This 
happens for no apparent reason (doesn't seem to have anything to do with 
any particular code or function).

Here is the last time:

 > library(help=ts)Killed

Has anyone seen this problem? Any idea on what I should fix/change?

Rick B.

"Richard A. Bilonick" <rab at nauticom.net> writes:

When I was running R 1.6.1 (under Red Hat 8.0), R would abruptly
terminate from time to time. I would type in a line of code and hit
return, and the cursor would not move from the end of the line. Then
the hard drive would start churning. This goes on for at least 10
minutes. Finally R terminates (finally freeing up the keyboard and
display). The only word printed is "Killed" immediately at the end of
the last line of code. I upgraded to 1.6.2 and the same thing
continues to happen. This happens for no apparent reason (doesn't seem
to have anything to do with any particular code or function).

Here is the last time:

 > library(help=ts)Killed

Has anyone seen this problem? Any idea on what I should fix/change?

Haven't heard about that before. There are three things you can do:

a) Upgrade to R 1.7.0 when it comes out "tomorrow" (for largish
values of tomorrow -- depends on what Martyn's plans are for
Easter...) and see if the problem is still there.

b) Next time it happens, run "top" or "ps" in another terminal, figure
out the process ID (say, 7913) and run strace -p 7913 and tell us what
you see.

c) Make a habit of starting R under the debugger: "R -d gdb", then in
gdb use "run". When the bad thing happens, Ctrl-C will bring you back
to the debugger where you say "bt" which might give us a clue. (If you
need to interrupt R for other reasons, use Ctrl-C and enter "signal
2"). You can also attach gdb to a running R much in the same way as
strace.

O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907

When I was running R 1.6.1 (under Red Hat 8.0), R would abruptly 
terminate from time to time. I would type in a line of code and hit 
return, and the cursor would not move from the end of the line. Then the 
hard drive would start churning. This goes on for at least 10 minutes. 
Finally R terminates (finally freeing up the keyboard and display). The 
only word printed is "Killed" immediately at the end of the last line of 
code. I upgraded to 1.6.2 and the same thing continues to happen. This 
happens for no apparent reason (doesn't seem to have anything to do with 
any particular code or function).

Here is the last time:

 > library(help=ts)Killed

Has anyone seen this problem? Any idea on what I should fix/change?

Can you reproduce this problem with "R --vanilla" - i.e. without loading
your saved workspace?

Do you get the same problem if you install the RPM for Red Hat 8.0?

What is happening here is that R is trying to allocate more and more
memory. Eventually it runs out of RAM and starts using the swap space on
your hard disk - hence the churning.  When the swap space is used up,
the operating system kills the offending process.

Martyn

Haven't heard about that before. There are three things you can do:

a) Upgrade to R 1.7.0 when it comes out "tomorrow" (for largish
values of tomorrow -- depends on what Martyn's plans are for
Easter...) and see if the problem is still there.

b) Next time it happens, run "top" or "ps" in another terminal, figure
out the process ID (say, 7913) and run strace -p 7913 and tell us what
you see.

c) Make a habit of starting R under the debugger: "R -d gdb", then in
gdb use "run". When the bad thing happens, Ctrl-C will bring you back
to the debugger where you say "bt" which might give us a clue. (If you
need to interrupt R for other reasons, use Ctrl-C and enter "signal
2"). You can also attach gdb to a running R much in the same way as
strace.


 

Thanks. I will try this. I have a habit of leaving many windows and 
programs open on 8 virtual desktops. I think this may be the cause of 
the slowdown in the system (as open windows and programs pile up). It 
takes longer and longer to open a new shell, etc. I closed all the 
programs and windows and then logged out and back in to see if that is 
connected to the problem running R.

I just have a few windows open and already it takes a LONG time to open 
a shell window. Other programs start right up. Guess it's time to go to 
RH help list.

Rick B.

Can you reproduce this problem with "R --vanilla" - i.e. without loading
your saved workspace?

Do you get the same problem if you install the RPM for Red Hat 8.0?

What is happening here is that R is trying to allocate more and more
memory. Eventually it runs out of RAM and starts using the swap space on
your hard disk - hence the churning.  When the swap space is used up,
the operating system kills the offending process.

Martyn



 

I haven't tried it with vanilla options. The problem occurs randomly and 
almost all my work is connected with one project at the moment. I 
installed both 1.6.1 and 1.6.2 from the rpm binary for RH 8.0. The 
Compaq Deskpro EN has  319mb of memory but this seems to be eaten up 
quickly with Mozilla running. Most of the physical memory gets cached 
right away.

Rick B.