Quantcast

Strange behaviour of Stata 11 under GNU/Linux

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Strange behaviour of Stata 11 under GNU/Linux

nshephard
Administrator
I was wondering if anyone else can replicate the following (strange)
behaviour of Stata 11.1 under GNU/Linux (or any other OS).

Hitting 'Ctrl + w' causes Stata to hang, using ~97-100% of one CPU and
has to be killed.

System Info
===========

. about

Stata/IC 11.1 for Unix (Linux 64-bit x86-64)
Born 16 Jun 2010
Copyright (C) 2009 StataCorp LP

Single-user Stata perpetual license:
       Serial number:  *******************
         Licensed to:  Neil Shephard
                       ScHARR, University of Sheffield

Kernel version...

$ uname -a
Linux morgan 2.6.34-gentoo-r2 #4 SMP Thu Jul 29 13:25:22 GMT 2010
x86_64 Intel(R) Core(TM)2 Duo CPU E8500 @ 3.16GHz GenuineIntel
GNU/Linux

GTK version...

$ eix -I gtk+
[I] x11-libs/gtk+
     Available versions:
        (1) 1.2.10-r12
        (2) 2.16.6 2.18.6 2.18.7 2.18.9 (~)2.20.1-r1
        {aqua cups debug doc +introspection jpeg jpeg2k linguas_az linguas_ca
linguas_cs linguas_da linguas_de linguas_el linguas_es linguas_et
linguas_eu linguas_fi linguas_fr linguas_ga linguas_gl linguas_hr
linguas_hu linguas_it linguas_ja linguas_ko linguas_lt linguas_nl
linguas_nn linguas_no linguas_pl linguas_pt linguas_pt_BR linguas_ro
linguas_ru linguas_sk linguas_sl linguas_sr linguas_sv linguas_tr
linguas_uk linguas_vi nls test tiff vim-syntax xinerama}
     Installed versions:  2.20.1-r1(2)(09:45:44 15/07/10)(cups jpeg
tiff -aqua -debug -doc -introspection -jpeg2k -test -vim-syntax
-xinerama)
     Homepage:            http://www.gtk.org/
     Description:         Gimp ToolKit +



Background
=========

'Ctrl + w' at a command prompt deletes back to the previous space in
the current command line.  I accidentally hit this whilst focused on
Stata as opposed to a terminal and immediately Stata 'hung' and the
CPU graph in the task bar went to 50% (NB Its a dual core system hence
the graph showed 50% usage as this is IC Stata as opposed to MP, a
check of -top- showed xstata to be using all the resources, see
below).

top - 09:52:11 up 20:06,  4 users,  load average: 0.43, 0.27, 0.17
Tasks: 144 total,   2 running, 141 sleeping,   0 stopped,   1 zombie
Cpu(s): 52.6%us,  1.0%sy,  0.0%ni, 46.4%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   2011168k total,  1623936k used,   387232k free,    46724k buffers
Swap:        0k total,        0k used,        0k free,   723056k cached


  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
14677 neil      20   0  284m  41m 6244 R   99  2.1   0:34.58 xstata
 2947 neil      20   0  175m  34m 6040 S    3  1.8  44:09.99 npviewer.bin
 2341 root      20   0  137m  28m 6976 S    1  1.4  68:19.84 X
 2401 mpd       20   0  218m 8560 2860 S    1  0.4   0:42.73 mpd
 2657 neil      20   0 1060m 442m  23m S    1 22.5  41:12.42 firefox
 2665 neil      20   0  130m 6804 5276 S    1  0.3   2:21.19 xfce4-cpugraph-
 2681 neil      20   0  316m  14m 8784 S    1  0.7   0:02.99 Terminal

Strange behaviour I thought, so before posting thought I'd see if it
was an artefact of the data that I had loaded/process I'd run by
starting a fresh session of xstata.  After a fresh start 'Ctrl + w'
had no effect, however, if I 'cd ~/tmp' and _then_ 'Ctrl + w' the
behaviour described above is replicated and Stata hangs.  No data was
loaded, nor any other commands used other than -cd- to change the
current working directory.

In typing this I have switched desktops and focus has moved away from
the Stata window, and after a few minutes the CPU usage drops back to
zero, but Stata is still hanging and no typing can be done.  If I
return to the desktop with Stata on and focus back on the window, CPU
usage goes back up to 97-100% maxing out one of the processors.

I thought I'd check to see if any more information could be obtained
when launching xstata from a terminal, but there was none.  The
behaviour is not observed in the CLI version either.

All very strange,

Neil

--
"... no scientific worker has a fixed level of significance at which
from year to year, and in all circumstances, he rejects hypotheses; he
rather gives his mind to each particular case in the light of his
evidence and his ideas." - Sir Ronald A. Fisher (1956)

Email - [hidden email]
Website - http://slack.ser.man.ac.uk/
Photos - http://www.flickr.com/photos/slackline/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange behaviour of Stata 11 under GNU/Linux

nshephard
Administrator
I checked this out on a colleagues Stata 11.1 under M$-Windows XP and
this behaviour could not be replicated.

Anyone else on a *NIX system (or OSX which is based on BSD) tested this out?

Neil




--
"... no scientific worker has a fixed level of significance at which
from year to year, and in all circumstances, he rejects hypotheses; he
rather gives his mind to each particular case in the light of his
evidence and his ideas." - Sir Ronald A. Fisher (1956)

Email - [hidden email]
Website - http://slack.ser.man.ac.uk/
Photos - http://www.flickr.com/photos/slackline/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

RE: Strange behaviour of Stata 11 under GNU/Linux

Tiago V. Pereira-2
In reply to this post by nshephard
Neil,

I got exactly the same behavior in Stata 10.1 (2.6.32-24-generic
#38-Ubuntu SMP Mon Jul 5 09:22:14 UTC 2010 i686 GNU/Linux), except that
xstata is not on 100% CPU Usage. Only pkill solves the problem, since
xstata is frozen. Very weird, indeed.



Tiago
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange behaviour of Stata 11 under GNU/Linux

Kevin H. Crow
In reply to this post by nshephard
Neil Shephard <[hidden email]> reported:

>I was wondering if anyone else can replicate the following (strange)
>behaviour of Stata 11.1 under GNU/Linux (or any other OS).
>
>Hitting 'Ctrl + w' causes Stata to hang, using ~97-100% of one CPU and
>has to be killed.

The keyboard shortcut ctrl-w in console Stata does a seek of the command
history (See [U] 10.4).  Stata for Unix(GUI) was also using this shortcut to
do a seek, but it should not have been.  In a future update, Stata for
Unix(GUI) will ignore this shortcut in the Command window and allow the
default key binding to take effect.  For other windows that support it, ctrl-w
will continue to close the current window.

Kevin
[hidden email]

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

RE: Strange behaviour of Stata 11 under GNU/Linux

Tiago V. Pereira-2
In reply to this post by nshephard
ariel,

based on what you want to do, the code you need is very simple. my may
question is, however, if you really need to create new weights each time
you bootstrap.

in any case, it is not very straightforward to write a code without seeing
the exact structure of your data and how exactly the weights are created.

sorry for the lack of caps.

tiago

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Strange behaviour of Stata 11 under GNU/Linux

nshephard
Administrator
In reply to this post by Kevin H. Crow
Kevin Crow wrote
Neil Shephard <nshephard@gmail.com> reported:

>I was wondering if anyone else can replicate the following (strange)
>behaviour of Stata 11.1 under GNU/Linux (or any other OS).
>
>Hitting 'Ctrl + w' causes Stata to hang, using ~97-100% of one CPU and
>has to be killed.

The keyboard shortcut ctrl-w in console Stata does a seek of the command
history (See [U] 10.4).  Stata for Unix(GUI) was also using this shortcut to
do a seek, but it should not have been.  In a future update, Stata for
Unix(GUI) will ignore this shortcut in the Command window and allow the
default key binding to take effect.  For other windows that support it, ctrl-w
will continue to close the current window.
Missed this at the end of last week (and clearly I'd missed [U] 10.4 too, oops), but good to hear it will be resolved in an update.

Cheers

Neil

Loading...