[R-SIG-Mac] mclapply THE_PROCESS_HAS_FORKED error

[Simon Urbanek]

On Jun 27, 2013, at 2:54 PM, Florian Oswald wrote:

> Hi Steve,
> 
> never tried the --vanilla approach, I guess i should. There are no plots in
> the code, but thanks for pointing that out, I'll watch out for this in
> future.
> 
> In some cases the function actually returned the correct value but R.app
> was spitting out the error message over and over again. in the activity
> monitor the 2 forked processes got stuck at 100% CPU usage. in other cases
> it crashed.
> 

Don't use the GUI - run R on the console to be safe. What R version is this? In recent versions we try to disable the event loop in forked children, but you may have an older versions or some package that still runs the event loop ...

Cheers,
Simon



> I'm sorry if the info I am giving is very generic. Any experience you can
> share is useful - thanks.
> 
> Florian
> 
> 
> On 27 June 2013 19:34, Steve Lianoglou <lianoglou.steve at gene.com> wrote:
> 
>> Hi,
>> 
>> On Thu, Jun 27, 2013 at 11:09 AM, Florian Oswald
>> <florian.oswald at gmail.com> wrote:
>>> Dear all,
>>> 
>>> I have got a question regarding the multicore package, used on a macbook
>>> pro with 4 cores. I have code that can be split into 2 independent
>> pieces,
>>> such that I call
>>> 
>>> result <- mclapply(1:2, function(x) myfun(x) )
>>> 
>>> It works perfectly - most of the time. Sometimes I get this error
>> message:
>>> 
>>> Break on
>>> 
>> __THE_PROCESS_HAS_FORKED_AND_YOU_CANNOT_USE_THIS_COREFOUNDATION_FUNCTIONALITY___YOU_MUST_EXEC__()
>>> to debug.
>>> 
>>> This happens particularly after I called mclapply over and over again in
>>> the same session. I have a suspicion that some forked processes are not
>>> properly terminated after the function exits (I can see them in the
>>> activity monitor). Notice I do not change option mc.cleanup (i.e. it's at
>>> default TRUE).
>>> 
>>> I imagine a source of error is myfun (too big to paste here) - the only
>>> thing I can say is that it runs without problem in lapply.
>>> 
>>> My biggest concern is that I want to run this function on a unix cluster.
>>> I'm not sure if what I'm saying makes a lot of sense but I want to make
>>> sure that I don't accumulate many idle R processes and run into that
>>> problem.
>> 
>> Does this problem only happen when you are running R through R.app? I
>> mean, if you fire up R from the terminal with `R --vanilla`, then try
>> your stuff again, do you see it happen?
>> 
>> Other things to look out for: is there plotting code in `myfun`?
>> 
>> -steve
>> 
>> --
>> Steve Lianoglou
>> Computational Biologist
>> Bioinformatics and Computational Biology
>> Genentech
>> 
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-SIG-Mac mailing list
> R-SIG-Mac at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-mac
> 
>