[slurm-users] Technical Update: Native Slurm Simulation Environment Successfully Configured

22 views
Skip to first unread message

sami sami via slurm-users

unread,
May 14, 2026, 5:52:03 AMMay 14
to slurm...@lists.schedmd.com

Hi All,

I am currently working on a single Linux machine running Ubuntu 25.10 and am looking to conduct performance testing on the Slurm backfill algorithm. My objective is to simulate a cluster environment consisting of 1,024 nodes.

While researching existing solutions, I found that many modern Slurm-simulator implementations (such as various plugin-based designs) rely heavily on Docker and Docker Compose. However, I would like to avoid using containerization and instead run the simulation natively on my host OS.

I understand that older versions of the Slurm source code included a native sim directory that allowed for simulations without external dependencies like Docker. Given this, I have the following technical questions:

  • Native Compatibility: Is it still feasible to compile and run a Slurm simulator natively on Ubuntu 25.10?

  • Version Recommendations: Would you recommend using a specific legacy version of Slurm that still contains the original simulation files, or is there a way to port that functionality to a more recent release?

  • Configuration: Are there specific build flags or configuration steps required to enable the simulated node environment (1,024 nodes) on a single machine without triggering the overhead of containerized networking?

I would appreciate any guidance or documentation you could provide on achieving a high-node-count simulation in a "bare-metal" Linux environment.

Best regards,

Patrick Pun via slurm-users

unread,
May 14, 2026, 5:20:05 PMMay 14
to slurm...@lists.schedmd.com
I think the option you are looking for is "--enable-multiple-slurmd" at build time, this should allow you start multiple slurmd instance on the same host (at different port) as if they are individual compute node.
https://slurm.schedmd.com/programmer_guide.html#multiple_slurmd_support

--
slurm-users mailing list -- slurm...@lists.schedmd.com
To unsubscribe send an email to slurm-us...@lists.schedmd.com

Manuel G. Marciani via slurm-users

unread,
May 28, 2026, 5:47:28 AM (6 days ago) May 28
to slurm...@lists.schedmd.com
Hello, Sami:

I am not sure if you are aware, but there is the Slurm simulator [1]
[2]. It has support for the backfill algorithm, which might help you in
your research.

Best,
Manu.


[1] https://dl.acm.org/doi/epdf/10.1145/3491418.3535178
[2] https://github.com/ubccr-slurm-simulator/slurm_simulator


On 14/5/26 11:44, sami sami via slurm-users wrote:
> Hi All,
>
> I am currently working on a single Linux machine running *Ubuntu 25.10*
> and am looking to conduct performance testing on the *Slurm backfill
> algorithm*. My objective is to simulate a cluster environment consisting
> of *1,024 nodes*.
>
> While researching existing solutions, I found that many modern
> Slurm-simulator implementations (such as various plugin-based designs)
> rely heavily on *Docker and Docker Compose*. However, I would like to
> avoid using containerization and instead run the simulation natively on
> my host OS.
>
> I understand that older versions of the Slurm source code included a
> native |sim| directory that allowed for simulations without external
> dependencies like Docker. Given this, I have the following technical
> questions:
>
> *
>
> *Native Compatibility:* Is it still feasible to compile and run a
> Slurm simulator natively on Ubuntu 25.10?
>
> *
>
> *Version Recommendations:* Would you recommend using a specific
> legacy version of Slurm that still contains the original simulation
> files, or is there a way to port that functionality to a more recent
> release?
>
> *
>
> *Configuration:* Are there specific build flags or configuration
> steps required to enable the simulated node environment (1,024
> nodes) on a single machine without triggering the overhead of
> containerized networking?
>
> I would appreciate any guidance or documentation you could provide on
> achieving a high-node-count simulation in a "bare-metal" Linux environment.
>
> Best regards,
>
>
>

--
Manuel G. Marciani - マルシアニ·マヌエル

First Stage Reseacher at Computational Earth Sciences (CES) - Earth
Sciences Department
Barcelona Supercomputing Center - Centro Nacional de Supercomputación

Ph.D Student at Departament d'Arquitectura de Computadors (DAC) -
Facultat d'Informàtica de Barcelona (FIB)
Universitat Politècnica de Catalunya (UPC)

BSC building
Plaça Eusebi Güell, 1-3, 08034 Barcelona, Spain
Desk BSC-PL0-6-20/8

mail to: manuel....@bsc.es
mail to: manuel.gimen...@upc.edu
mail to: manuel....@a.riken.jp
mail to: manuel.gimenez.d...@hu-berlin.de
Reply all
Reply to author
Forward
0 new messages