[lustre-discuss] lustre-discuss Digest, Vol 191, Issue 2

Thu Feb 3 08:41:26 PST 2022

Hi John,

Thanks for looking at the code.

Right for the loop, but this clearly will not affect the performance on IO.

About concurrent write jobs: i am experiencing very low rate even with only

1 MPI process ( so no concurrency at all ).

About the array size not being a multiple of stripe size, this is affecting the performance

in case i have many MPI  writing process,  right ?

Best,

Denis

________________________________
From: lustre-discuss <lustre-discuss-bounces at lists.lustre.org> on behalf of John Bauer <bauerj at iodoctors.com>
Sent: Thursday, February 3, 2022 5:16:52 PM
To: lustre-discuss at lists.lustre.org
Subject: Re: [lustre-discuss] lustre-discuss Digest, Vol 191, Issue 2

 The following loop in wdfile.f90 is pointless as the write happens only once for each rank.

Each rank is writing out the array once and then closing the file.  If the size of array 'data' is not a multiple of the Lustre stripe size there is going to be a lot of read-modify-write going on.

 do ii = 0, size
     if ( rank == ii ) then
        !start= MPI_Wtime()
        write(unit=iounit) data(1:nx, 1:ny, 1:nz)
        close(iounit)
        !finish = MPI_Wtime()
        !write(6,'(i5,f7.4)') rank, finish - start
     else
     end if
  end do

On 2/3/2022 9:39 AM, lustre-discuss-request at lists.lustre.org<mailto:lustre-discuss-request at lists.lustre.org> wrote:

Send lustre-discuss mailing list submissions to
        lustre-discuss at lists.lustre.org<mailto:lustre-discuss at lists.lustre.org>

To subscribe or unsubscribe via the World Wide Web, visit
        http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
or, via email, send a message with subject or body 'help' to
        lustre-discuss-request at lists.lustre.org<mailto:lustre-discuss-request at lists.lustre.org>

You can reach the person managing the list at
        lustre-discuss-owner at lists.lustre.org<mailto:lustre-discuss-owner at lists.lustre.org>

When replying, please edit your Subject line so it is more specific
than "Re: Contents of lustre-discuss digest..."

Today's Topics:

   1. RE-Fortran IO problem (Bertini, Denis Dr.)
   2. Re: RE-Fortran IO problem (Patrick Farrell)
   3. Re: RE-Fortran IO problem (Bertini, Denis Dr.)

----------------------------------------------------------------------

Message: 1
Date: Thu, 3 Feb 2022 12:43:21 +0000
From: "Bertini, Denis Dr." <D.Bertini at gsi.de><mailto:D.Bertini at gsi.de>
To: "lustre-discuss at lists.lustre.org"<mailto:lustre-discuss at lists.lustre.org>
        <lustre-discuss at lists.lustre.org><mailto:lustre-discuss at lists.lustre.org>
Subject: [lustre-discuss] RE-Fortran IO problem
Message-ID: <b38ed90e7586428f9f46f742b0b359ef at gsi.de><mailto:b38ed90e7586428f9f46f742b0b359ef at gsi.de>
Content-Type: text/plain; charset="iso-8859-1"

Hi,

Just as an add-on to my previous mail, the problem shows up also

with intel fortran  and it not specific to gnu fortran compiler.

So it seems to be linked to how the fortran IO is handled which

seems to be sub-optimal in cas of a Lustre filesystem.

I would be grateful if one can confirm/disconfirm  that.

Here again the access to the code i used for my benchmarks:

https://git.gsi.de/hpc/cluster/ci_ompi/-/tree/main/f/src

Best,

Denis

---------
Denis Bertini
Abteilung: CIT
Ort: SB3 2.265a

Tel: +49 6159 71 2240
Fax: +49 6159 71 2986
E-Mail: d.bertini at gsi.de<mailto:d.bertini at gsi.de>

GSI Helmholtzzentrum f?r Schwerionenforschung GmbH
Planckstra?e 1, 64291 Darmstadt, Germany, www.gsi.de<http://www.gsi.de>

Commercial Register / Handelsregister: Amtsgericht Darmstadt, HRB 1528
Managing Directors / Gesch?ftsf?hrung:
Professor Dr. Paolo Giubellino, Dr. Ulrich Breuer, J?rg Blaurock
Chairman of the GSI Supervisory Board / Vorsitzender des GSI-Aufsichtsrats:
Ministerialdirigent Dr. Volkmar Dietz
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20220203/9f68dae9/attachment-0001.html><http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20220203/9f68dae9/attachment-0001.html>

------------------------------

Message: 2
Date: Thu, 3 Feb 2022 15:15:16 +0000
From: Patrick Farrell <pfarrell at ddn.com><mailto:pfarrell at ddn.com>
To: "Bertini, Denis Dr." <D.Bertini at gsi.de><mailto:D.Bertini at gsi.de>,
        "lustre-discuss at lists.lustre.org"<mailto:lustre-discuss at lists.lustre.org> <lustre-discuss at lists.lustre.org><mailto:lustre-discuss at lists.lustre.org>
Subject: Re: [lustre-discuss] RE-Fortran IO problem
Message-ID:
        <DM6PR19MB3129DB8B22167BBDB55569B3C9289 at DM6PR19MB3129.namprd19.prod.outlook.com><mailto:DM6PR19MB3129DB8B22167BBDB55569B3C9289 at DM6PR19MB3129.namprd19.prod.outlook.com>

Content-Type: text/plain; charset="utf-8"

Denis,

FYI, the git link you provided seems to be non-public - it asks for a GSI login.

Fortran is widely used for applications on Lustre, so it's unlikely to be a fortran specific issue.  If you're seeing I/O rates drop suddenly during? activity, rather than being reliably low for some particular operation, I would look to the broader Lustre system.  It may be suddenly extremely busy or there could be, eg, a temporary network issue - Assuming this is a system belonging to your institution, I'd check with your admins.

Regards,
Patrick
________________________________
From: lustre-discuss <lustre-discuss-bounces at lists.lustre.org><mailto:lustre-discuss-bounces at lists.lustre.org> on behalf of Bertini, Denis Dr. <D.Bertini at gsi.de><mailto:D.Bertini at gsi.de>
Sent: Thursday, February 3, 2022 6:43 AM
To: lustre-discuss at lists.lustre.org<mailto:lustre-discuss at lists.lustre.org> <lustre-discuss at lists.lustre.org><mailto:lustre-discuss at lists.lustre.org>
Subject: [lustre-discuss] RE-Fortran IO problem

Hi,

Just as an add-on to my previous mail, the problem shows up also

with intel fortran  and it not specific to gnu fortran compiler.

So it seems to be linked to how the fortran IO is handled which

seems to be sub-optimal in cas of a Lustre filesystem.

I would be grateful if one can confirm/disconfirm  that.

Here again the access to the code i used for my benchmarks:

https://git.gsi.de/hpc/cluster/ci_ompi/-/tree/main/f/src

Best,

Denis

---------
Denis Bertini
Abteilung: CIT
Ort: SB3 2.265a

Tel: +49 6159 71 2240
Fax: +49 6159 71 2986
E-Mail: d.bertini at gsi.de<mailto:d.bertini at gsi.de>

GSI Helmholtzzentrum f?r Schwerionenforschung GmbH
Planckstra?e 1, 64291 Darmstadt, Germany, www.gsi.de<http://www.gsi.de>

Commercial Register / Handelsregister: Amtsgericht Darmstadt, HRB 1528
Managing Directors / Gesch?ftsf?hrung:
Professor Dr. Paolo Giubellino, Dr. Ulrich Breuer, J?rg Blaurock
Chairman of the GSI Supervisory Board / Vorsitzender des GSI-Aufsichtsrats:
Ministerialdirigent Dr. Volkmar Dietz
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20220203/87d50e4f/attachment-0001.html><http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20220203/87d50e4f/attachment-0001.html>

------------------------------

Message: 3
Date: Thu, 3 Feb 2022 15:38:58 +0000
From: "Bertini, Denis Dr." <D.Bertini at gsi.de><mailto:D.Bertini at gsi.de>
To: Patrick Farrell <pfarrell at ddn.com><mailto:pfarrell at ddn.com>,
        "lustre-discuss at lists.lustre.org"<mailto:lustre-discuss at lists.lustre.org> <lustre-discuss at lists.lustre.org><mailto:lustre-discuss at lists.lustre.org>
Subject: Re: [lustre-discuss] RE-Fortran IO problem
Message-ID: <9476216bb77b4905b0ee8db9598dc6df at gsi.de><mailto:9476216bb77b4905b0ee8db9598dc6df at gsi.de>
Content-Type: text/plain; charset="utf-8"

Dear Patrick,

Thanks for the quick answer.

Sorry for the broken link, i send anyway the code as an tarball in

my previous mail

I am myself using fortran together with Lustre,  using a different application

IO than pure fortran i.e MPI-IO or HDF5 and it works just fine.

I never used the pure fortran IO with lustre and was surprised by the

low performance on our filesystem that users reported.

In the tarball i adapted the pure fortran code to use a different application IO (HDF5).

In this case i see no preformance problem.

Could it be that i both case, there are data contention problem and/or temporal network

issue but the HDF5 IO is just more resilient  than fortran IO to these issue.

Anyway how to check properly for these Lustre system problem ( data contention/network)?

Regards,

Denis

________________________________
From: Patrick Farrell <pfarrell at ddn.com><mailto:pfarrell at ddn.com>
Sent: Thursday, February 3, 2022 4:15:16 PM
To: Bertini, Denis Dr.; lustre-discuss at lists.lustre.org<mailto:lustre-discuss at lists.lustre.org>
Subject: Re: RE-Fortran IO problem

Denis,

FYI, the git link you provided seems to be non-public - it asks for a GSI login.

Fortran is widely used for applications on Lustre, so it's unlikely to be a fortran specific issue.  If you're seeing I/O rates drop suddenly during? activity, rather than being reliably low for some particular operation, I would look to the broader Lustre system.  It may be suddenly extremely busy or there could be, eg, a temporary network issue - Assuming this is a system belonging to your institution, I'd check with your admins.

Regards,
Patrick
________________________________
From: lustre-discuss <lustre-discuss-bounces at lists.lustre.org><mailto:lustre-discuss-bounces at lists.lustre.org> on behalf of Bertini, Denis Dr. <D.Bertini at gsi.de><mailto:D.Bertini at gsi.de>
Sent: Thursday, February 3, 2022 6:43 AM
To: lustre-discuss at lists.lustre.org<mailto:lustre-discuss at lists.lustre.org> <lustre-discuss at lists.lustre.org><mailto:lustre-discuss at lists.lustre.org>
Subject: [lustre-discuss] RE-Fortran IO problem

Hi,

Just as an add-on to my previous mail, the problem shows up also

with intel fortran  and it not specific to gnu fortran compiler.

So it seems to be linked to how the fortran IO is handled which

seems to be sub-optimal in cas of a Lustre filesystem.

I would be grateful if one can confirm/disconfirm  that.

Here again the access to the code i used for my benchmarks:

https://git.gsi.de/hpc/cluster/ci_ompi/-/tree/main/f/src

Best,

Denis

---------
Denis Bertini
Abteilung: CIT
Ort: SB3 2.265a

Tel: +49 6159 71 2240
Fax: +49 6159 71 2986
E-Mail: d.bertini at gsi.de<mailto:d.bertini at gsi.de>

GSI Helmholtzzentrum f?r Schwerionenforschung GmbH
Planckstra?e 1, 64291 Darmstadt, Germany, www.gsi.de<http://www.gsi.de>

Commercial Register / Handelsregister: Amtsgericht Darmstadt, HRB 1528
Managing Directors / Gesch?ftsf?hrung:
Professor Dr. Paolo Giubellino, Dr. Ulrich Breuer, J?rg Blaurock
Chairman of the GSI Supervisory Board / Vorsitzender des GSI-Aufsichtsrats:
Ministerialdirigent Dr. Volkmar Dietz
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20220203/10fb4b81/attachment.html><http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20220203/10fb4b81/attachment.html>

------------------------------

Subject: Digest Footer

_______________________________________________
lustre-discuss mailing list
lustre-discuss at lists.lustre.org<mailto:lustre-discuss at lists.lustre.org>
http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org

------------------------------

End of lustre-discuss Digest, Vol 191, Issue 2
**********************************************

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20220203/e1520ad9/attachment.html>