[Rd] parallel::detectCores(TRUE) gives: Error in system(cmd, TRUE) : error in running command

William Dunlap wdunlap at tibco.com
Sat Aug 23 00:57:35 CEST 2014


There is no /usr/sbin/sysctl on my Ubuntu 12.04 machine.  There is a
/sbin/sysctl, but it does not accept the '-n hw.ncpu' arguments.  Its
/usr/bin/nproc [-all] will tell the number of available [installed]
processing units and 'cat /proc/cpuinfo | grep processor | wc -l' will
also give the number of installed processing units.
Bill Dunlap
TIBCO Software
wdunlap tibco.com


On Fri, Aug 22, 2014 at 3:17 PM, Marius Hofert
<marius.hofert at uwaterloo.ca> wrote:
> Thanks, Bill. The output is:
>
>> trace(system, quote(print(command)))
>   parallel::detectCores(TRUE)
> Tracing function "system" in package "base"
> [1] "system"
>>
> Tracing system(cmd, TRUE) on entry
> [1] "/usr/sbin/sysctl -n hw.ncpu 2>/dev/null"
> Error in system(cmd, TRUE) : error in running command
>>
>
> On Fri, Aug 22, 2014 at 6:13 PM, William Dunlap <wdunlap at tibco.com> wrote:
>> The same is true in R-2.14.1 on  Ubuntu 12.04.4 LTS .  Put a trace on
>> system with
>>   trace(system, quote(print(command)))
>>   parallel::detectCores(TRUE)
>> and you will see the offending shell command.
>> Bill Dunlap
>> TIBCO Software
>> wdunlap tibco.com
>>
>>
>> On Fri, Aug 22, 2014 at 1:03 PM, Marius Hofert
>> <marius.hofert at uwaterloo.ca> wrote:
>>> Hi,
>>>
>>> Both under the current R-devel (r66456) and a version from about 3
>>> months ago, I experience the following behavior:
>>>
>>>> parallel::detectCores(TRUE)
>>> Error in system(cmd, TRUE) : error in running command
>>>> traceback()
>>> 3: system(cmd, TRUE)
>>> 2: gsub("^ +", "", system(cmd, TRUE)[1])
>>> 1: parallel::detectCores(TRUE)
>>>>
>>>
>>> This is on Ubuntu 14.04. Does anybody else see this? [I currently have
>>> quite a heavy workload, otherwise I would have installed and tested it
>>> also under 3.1.1]
>>>
>>> Cheers,
>>>
>>> Marius
>>>
>>> ______________________________________________
>>> R-devel at r-project.org mailing list
>>> https://stat.ethz.ch/mailman/listinfo/r-devel



More information about the R-devel mailing list