caml-list - the Caml user's mailing list
 help / color / mirror / Atom feed
From: Anders Fugmann <anders@fugmann.net>
To: Gerd Stolpmann <info@gerd-stolpmann.de>
Cc: caml-list <caml-list@inria.fr>
Subject: Re: [Caml-list] Netmcore problems.
Date: Mon, 16 Jan 2012 18:46:53 +0100	[thread overview]
Message-ID: <4F14628D.60809@fugmann.net> (raw)
In-Reply-To: <1326474410.14288.116.camel@thinkpad>

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

Thank you very much. Remounting with size=80% solved the problem.

It would seem that on debian, /run/shm (which is symlinked from
/dev/shm) is mounted with size=20%.. Quite conservative it would seem.

Again, thanks.
Anders

- --
Anders Fugmann

On 2012-01-13 18:06, Gerd Stolpmann wrote:
> Am Freitag, den 13.01.2012, 17:30 +0100 schrieb Anders Fugmann:
>> Hi,
>>
>> I'm having a problem using netmcore from ocamlnet 3.4 library.
>>
>> When I try to place a large data structure into Netmcore_array, the 
>> controlling process dies with the following error:
>>
>> [Fri Jan 13 16:21:38 2012] [netplex.controller] [alert] Process 28067 
>> for service netmcore_0 terminated with signal 7
> 
> That's sigbus. You get it when the OS does not have memory anymore - but
> not when the memory is allocated, but first when it is filled with data.
> You can change this by disallowing that memory is overcommitted (set the
> kernel param vm.overcommit_memory to 1 or even 2).
> 
>> The error occurs while trying to initialise a Netmcore_array from an 
>> array of 4*10^6 strings all of length 150. This is done in the context 
>> of the "first" process.
>>
>> The controlling process initializes a memory pool by using
>>
>>    let pool_id = Netmcore_mempool.create_mempool (1024 * 1024 * 1024 * 
>> 10) in
>>
>> Running the code on a smaller dataset seems to work.
>>
>> I have increased kernel variables shmall and shmmax to very high numbers 
>> (100Gb), but it does not solve the problem.
> 
> No, this cannot solve it. These variables only control System V shared
> memory, but Netmcore uses POSIX shared memory. On Linux, you can change
> the max of this memory by re-mounting /dev/shm, e.g.
> 
> mount -o remount,size=80% /dev/shm
> 
> The default is 50% of available RAM.
> 
>> All ideas are welcome.
>>
>> On other question while I'm at it; am I allowed to create multiple 
>> shared arrays from the same memory pool, or do I need to create one pool 
>> for each shared array?
> 
> The pools can be shared.
> 
> Gerd
> 
>>
>> Regards
>> Anders Fugmann
>>
>>
>>
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk8UYo0ACgkQ9GexfkaZ7hSfLwCgiu+grFYsZXyQlvvnUBZgna+4
twkAoJKyNJ0hXlI0u9IqmuapqQ48YGDz
=RCbL
-----END PGP SIGNATURE-----

      reply	other threads:[~2012-01-16 17:46 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-01-13 16:30 Anders Fugmann
2012-01-13 17:06 ` Gerd Stolpmann
2012-01-16 17:46   ` Anders Fugmann [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4F14628D.60809@fugmann.net \
    --to=anders@fugmann.net \
    --cc=caml-list@inria.fr \
    --cc=info@gerd-stolpmann.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).