I too have experienced issues when having too many packages installed.
(beyond a fun exercise of curiosity, I don't recommend yum install * ;) ).
I may try out the rpmbuild idea you have though, Jacco.

I have a single file in /etc/security/limits.d called 20-nproc.conf.
Contents of this file are as follows:

# Default limit for number of user's processes to prevent
# accidental fork bombs.
# See rhbz #432903 for reasoning.

*          soft    nproc     4096
root       soft    nproc     unlimited

So it appears it's set for 4MB.  Is that an acceptable number?  Or should I
lower it?

On Fri, Apr 17, 2015 at 5:23 AM, Gordan Bobic <[email protected]> wrote:

> It might be worth checking what is in your /etc/security/limits.d
> just in case. Reducing the default stack allocation size is a
> good way to reduce the memory footprint of a running system,
> and all programs that aren't broken should be able to
> request (and receive) extra stack space if they require it.
>
> In reality, it doesn't always work out like that (e.g. yum
> is known to break before most things when you reduce the
> stack size dramatically).
>
> IIRC, default stack size on EL6 is 8MB, but the vast
> majority of programs work just fine with 256KB.
>
> Gordan
>
>
> On 2015-04-17 02:30, Mark Campbell wrote:
>
>> Gordon,
>>
>> I haven't done anything with either of those.  Beyond it starting out
>> as a bare bone RSEL 7 image, me installing the RPi2 kernel from RSEL
>> 6, and then me installing packages via yum as I found the build
>> failing due to missing packages, I've done virtually no customization
>> to the environment.  That's why I figured there might've been some
>> system variable that may have been affecting this somehow.
>>
>> Perhaps the issue is still one of me not having the necessary packages
>> installed, it's just hiding under a misleading error?  Jacco, would it
>> be terribly inconvenient if I could get a list of packages you have
>> installed on your rpi2?  Maybe if I replicate what you have installed,
>> my problem will go away?
>>
>> On Thursday, April 16, 2015, Gordan Bobic <[email protected]>
>> wrote:
>>
>>  Mark,
>>>
>>> Just a thoughts - did you perchance reduce the default stack
>>> allocation using ulimit or limits.d?
>>>
>>> Gordan
>>>
>>> On 2015-04-16 22:00, Jacco Ligthart wrote:
>>> Hi Mark,
>>>
>>> I cannot reproduce your error.
>>> I tried on both RSEL7 on odroid and RSEL7.1 on raspberrypi2.
>>>
>>> Both times I did:
>>> cd /tmp/
>>> tar -xzvf mpich-3.1.4.tar.gz
>>> cd mpich-3.1.4
>>> ./configure
>>> make
>>>
>>> and also both times it seems to complete building without error.
>>>
>>> Jacco
>>>
>>> On 04/13/2015 11:23 PM, Mark Campbell wrote:
>>>
>>> Thanks Bjarne,
>>>
>>> Google is always the first place I try, but I was less than
>>> successful with this one. I'd seen numerous people reporting the
>>> same thing as me, and I saw a variety of answers, some not
>>> applicable to me, others would suggest that somehow the arguments
>>> would have to be broken up (not really sure how you would do that
>>> with Make), others still would suggest reducing the path
>>> environment
>>> variable, others still suggested that the max command length would
>>> need to be extended (haven't figured out how that is done yet, if
>>> anyone does, I'm all ears), and the rest were unresolved.
>>>
>>> I tried moving it to a much shorter directory-- /mpi, and that
>>> still
>>> had problems.
>>>
>>> I can't really tell you what it's listing that's too long--I don't
>>> see it myself. All I do is:
>>>
>>> # ./configure
>>> # make
>>>
>>> ./configure goes through what must be 1000+ lines of checking, with
>>> no errors that I could see, and make comes back right at the start
>>> with:
>>>
>>> make[1]: execvp: /bin/sh: Argument list too long
>>>
>>> And something about error 27, with a general make error dialog. (I
>>> don't have the exact dialog in front of me right now, the pi is at
>>> home, and currently off).
>>>
>>> If it were just a distribution build dependency problem, wouldn't
>>> the problem show up when doing the ./configure script, where it
>>> checks for all of the dependencies?
>>>
>>> On Mon, Apr 13, 2015 at 4:24 PM, Bjarne Saltbæk
>>> <[email protected]> wrote:
>>>
>>> Google knows everything... :)
>>>
>>> https://gcc.gnu.org/ml/gcc-help/2010-12/msg00305.html [1] [1]
>>>
>>> could be a path issue.
>>> could be a distribution build dependency issue.
>>> maybe you can post what list that is to long
>>>
>>> BR,
>>> Bjarne
>>>
>>> -------------------------
>>> Date: Sun, 12 Apr 2015 23:24:33 -0400
>>> From: [email protected]
>>> To: [email protected]
>>> Subject: [RedSleeve-Users] RSEL7 Make issue
>>>
>>> So I am trying to compile mpich on RSEL7 (I know the RPMs exist in
>>> yum, and I've even installed them, but the rpms don't appear to
>>> come with the examples code), and I am running into a problem
>>> building it.
>>>
>>> After I've successfully run the ./configure file with no reported
>>> errors, I tell it to make, and make complains that the argument
>>> list is too long (actual error is execvp: /bin/sh: Argument list
>>> too long). Using this same source, I was able to make/make
>>> install fine on Raspbian.
>>>
>>> Any ideas on why it won't build here?
>>>
>>> --
>>>
>>> --Mark
>>> _______________________________________________ users mailing
>>> list [email protected]
>>> http://lists.redsleeve.org/mailman/listinfo/users [2] [2]
>>> _______________________________________________
>>> users mailing list
>>> [email protected]
>>> http://lists.redsleeve.org/mailman/listinfo/users [2] [2]
>>>
>>> --
>>>
>>> --Mark
>>>
>>> _______________________________________________
>>> users mailing list
>>> [email protected]
>>> http://lists.redsleeve.org/mailman/listinfo/users [2] [2]
>>>
>>
>>  Links:
>>  ------
>>  [1] https://gcc.gnu.org/ml/gcc-help/2010-12/msg00305.html [1]
>>  [2] http://lists.redsleeve.org/mailman/listinfo/users [2]
>>
>>  _______________________________________________
>>  users mailing list
>>  [email protected]
>>  http://lists.redsleeve.org/mailman/listinfo/users [2]
>>  _______________________________________________
>>  users mailing list
>>  [email protected]
>>  http://lists.redsleeve.org/mailman/listinfo/users [2]
>>
>> --
>> --Mark
>>
>>
>> Links:
>> ------
>> [1] https://gcc.gnu.org/ml/gcc-help/2010-12/msg00305.html
>> [2] http://lists.redsleeve.org/mailman/listinfo/users
>>
>> _______________________________________________
>> users mailing list
>> [email protected]
>> http://lists.redsleeve.org/mailman/listinfo/users
>>
> _______________________________________________
> users mailing list
> [email protected]
> http://lists.redsleeve.org/mailman/listinfo/users
>



-- 
--Mark
_______________________________________________
users mailing list
[email protected]
http://lists.redsleeve.org/mailman/listinfo/users

Reply via email to