Skip to content

R process gets killed spontaneously

8 messages · Peng Yu, Uwe Ligges, Benilton Carvalho +2 more

#
My R process has been killed for a few times, although the system
administrator did not do so. It happened when R attempted to allocate
a lot of memory. I'm wondering whether R would spontaneously kill
itself if it can not allocate enough memory?
#
Peng Yu wrote:
It does not kill itself. If it was not some out-of-memory kill process 
of your OS, then you may have seen some segfault. We need reproducible 
code and all the details described in the posting guide.

Uwe Ligges
#
2009/11/10 Uwe Ligges <ligges at statistik.tu-dortmund.de>:
Please see this thread about the necessary information.

http://www.mail-archive.com/r-help at r-project.org/msg74804.html
#
Hi Peng,

in a very simplistic manner, what happens is that the Operating System  
"thinks" it is too dangerous to let the R process to use so much  
memory. So, to protect the whole system, it kills R, before the system  
becomes unstable.

I've been looking at the problem you observed last week and, as I  
promised, I will come back to you with my findings sometime this week.  
I was already informed on how to reduce the memory footprint when  
creating the specific object you're interested in, but still need to  
run some tests.

Will get back to you soon,

b
On Nov 10, 2009, at 2:23 PM, Peng Yu wrote:

            
#
This was happening to me on Red Hat Linux when I was running huge jobs within a screen session. By any chance are your R processes running within a screen session? (screen is a very nice program that will keep your sessions alive after you log out, but it was killing off my big memory jobs for unknown reasons.)

Eric



----- Original message -----
From: "Peng Yu" <pengyu.ut at gmail.com>
To: r-help at stat.math.ethz.ch
Date: Tue, 10 Nov 2009 10:23:18 -0600
Subject: Re: [R] R process gets killed spontaneously

2009/11/10 Uwe Ligges <ligges at statistik.tu-dortmund.de>:
Please see this thread about the necessary information.

http://www.mail-archive.com/r-help at r-project.org/msg74804.html

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
#
I run R in gnome-terminal. Is it what you referred as 'screen session'?
On Tue, Nov 10, 2009 at 3:01 PM, <rmailbox at justemail.net> wrote:
#
I think he was referring to the actual 'screen' command..

but, I digress... Is there any way you can put your commands in a script 
and execute them from the command line so that you see the actual memory 
and GC output from R in realtime? I use the following (in WinXP) to 
debug any faulty processes or detect if the job was actually completed 
successfully:

(adapted for Linux, *not* winxp)
R --verbose MorningStartup.r > morningsummary-log.txt

That should yield some more results which hopefully should help you 
investigate. The pertinent stuff is output to the console.

-cedrick
Peng Yu wrote:
#
Cedrick is correct. I was referring to the command invoked by entering the string "screen" at the shell prompt.


----- Original message -----
From: "Cedrick W. Johnson" <cedrick at cedrickjohnson.com>
To: "Peng Yu" <pengyu.ut at gmail.com>
Cc: r-help at r-project.org
Date: Tue, 10 Nov 2009 16:38:09 -0500
Subject: Re: [R] R process gets killed spontaneously

I think he was referring to the actual 'screen' command..

but, I digress... Is there any way you can put your commands in a script 
and execute them from the command line so that you see the actual memory 
and GC output from R in realtime? I use the following (in WinXP) to 
debug any faulty processes or detect if the job was actually completed 
successfully:

(adapted for Linux, *not* winxp)
R --verbose MorningStartup.r > morningsummary-log.txt

That should yield some more results which hopefully should help you 
investigate. The pertinent stuff is output to the console.

-cedrick
Peng Yu wrote:
______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.