"Requested node configuration is not available" when using -c

[Michal Zielinski]

I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with
CR_CPU and no usage restrictions. Each node has just 1 CPU.

Here's what I can't figure out:

Running* srun -c 1 hostname* works, and it returns *node1*.

Running *srun -c 1 -x node1 hostname* also works, and it returns *node2*.

But, *srun -c 2 hostname* does not work, and it returns the above error.

I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

Attached is my slurm.conf.

Thanks for your help.

[Christopher Samuel]

Post by Michal Zielinski
I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with
CR_CPU and no usage restrictions. Each node has just 1 CPU.

[...]

Post by Michal Zielinski
But, *srun -c 2 hostname* does not work, and it returns the above error.
I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?

cheers,
Chris

[Josh McSavaney]

I believe your slurm.conf is defining 4 nodes with a single logical
processor each. You are then trying to allocate two CPUs on a single node
with srun, which (according to your slurm.conf) you do not have.

You may want to consider `srun -n 2 hostname` and see where that lands you.

Regards,

Josh McSavaney
Bit Flipper
Rochester Institute of Technology

Post by Michal Zielinski
I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with

Post by Michal Zielinski
CR_CPU and no usage restrictions. Each node has just 1 CPU.

[...]

Post by Michal Zielinski
But, *srun -c 2 hostname* does not work, and it returns the above error.
I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?
cheers,
Chris
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
http://www.vlsci.org.au/ http://twitter.com/vlsci

[Michal Zielinski]

Josh,

I believe that *-n *sets the number of tasks. I only want a single task, as
when a single process uses multiple cores. *srun -n 2 hostname* returns

linux-slurm2
linux-slurm3

which is definitely not what I want.

Thanks,
Mike

Post by Josh McSavaney
I believe your slurm.conf is defining 4 nodes with a single logical
processor each. You are then trying to allocate two CPUs on a single node
with srun, which (according to your slurm.conf) you do not have.
You may want to consider `srun -n 2 hostname` and see where that lands you.
Regards,
Josh McSavaney
Bit Flipper
Rochester Institute of Technology

Post by Michal Zielinski
I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with

Post by Michal Zielinski
CR_CPU and no usage restrictions. Each node has just 1 CPU.

[...]

Post by Michal Zielinski
But, *srun -c 2 hostname* does not work, and it returns the above error.
I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?
cheers,
Chris
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
http://www.vlsci.org.au/ http://twitter.com/vlsci

[Eckert, Phil]

Mike,

In your slurm.conf you have Procs=1, (which is the same as CPUS=1) and Sockets (if ommited will be inferred from CPUS, default is 1) and CoresPerSocket (default is 1)

So at this point the slurm.conf has a default configuration of 1 core per node.

Phil Eckert
LLNL

From: Michal Zielinski <michal.zielinski-jZvHcCJSbXyHXe+***@public.gmane.org<mailto:***@uconn.edu>>
Reply-To: slurm-dev <slurm-dev-***@public.gmane.org<mailto:slurm-dev-***@public.gmane.org>>
Date: Tuesday, September 9, 2014 at 6:35 AM
To: slurm-dev <slurm-dev-***@public.gmane.org<mailto:slurm-dev-***@public.gmane.org>>
Subject: [slurm-dev] Re: "Requested node configuration is not available" when using -c

Josh,

I believe that -n sets the number of tasks. I only want a single task, as when a single process uses multiple cores. srun -n 2 hostname returns

linux-slurm2
linux-slurm3

which is definitely not what I want.

Thanks,
Mike


On Mon, Sep 8, 2014 at 8:07 PM, Josh McSavaney <McSaucy-s/YdZray+***@public.gmane.org<mailto:McSaucy-s/YdZray+***@public.gmane.org>> wrote:
I believe your slurm.conf is defining 4 nodes with a single logical processor each. You are then trying to allocate two CPUs on a single node with srun, which (according to your slurm.conf) you do not have.

You may want to consider `srun -n 2 hostname` and see where that lands you.

Regards,

Josh McSavaney
Bit Flipper
Rochester Institute of Technology



On Mon, Sep 8, 2014 at 7:42 PM, Christopher Samuel <samuel-***@public.gmane.org<mailto:samuel-***@public.gmane.org>> wrote:

On 09/09/14 07:26, Michal Zielinski wrote:

I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with
CR_CPU and no usage restrictions. Each node has just 1 CPU.
[...]
But, *srun -c 2 hostname* does not work, and it returns the above error.

I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?

cheers,
Chris
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
Email: samuel-***@public.gmane.org<mailto:samuel-***@public.gmane.org> Phone: [phone]<tel:%2B61%20%280%293%20903%2055545>
http://www.vlsci.org.au/ http://twitter.com/vlsci

[Michal Zielinski]

Phil,

I believe that 1 core per node is correct.

Maybe let me ask this first: is possible for a single task to use specific
CPUs across several nodes?

Thanks,
Mike

Post by Eckert, Phil
Mike,
In your slurm.conf you have Procs=1, (which is the same as CPUS=1) and
Sockets (if ommited will be inferred from CPUS, default is 1) and
CoresPerSocket (default is 1)
So at this point the slurm.conf has a default configuration of 1 core per node.
Phil Eckert
LLNL
Date: Tuesday, September 9, 2014 at 6:35 AM
Subject: [slurm-dev] Re: "Requested node configuration is not available" when using -c
Josh,
I believe that *-n *sets the number of tasks. I only want a single task,
as when a single process uses multiple cores. *srun -n 2 hostname* returns
linux-slurm2
linux-slurm3
which is definitely not what I want.
Thanks,
Mike

Post by Josh McSavaney
I believe your slurm.conf is defining 4 nodes with a single logical
processor each. You are then trying to allocate two CPUs on a single node
with srun, which (according to your slurm.conf) you do not have.
You may want to consider `srun -n 2 hostname` and see where that lands you.
Regards,
Josh McSavaney
Bit Flipper
Rochester Institute of Technology

Post by Michal Zielinski
I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with

Post by Michal Zielinski
CR_CPU and no usage restrictions. Each node has just 1 CPU.

[...]

Post by Michal Zielinski
But, *srun -c 2 hostname* does not work, and it returns the above error.
I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?
cheers,
Chris
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
http://www.vlsci.org.au/ http://twitter.com/vlsci

[Chrysovalantis Paschoulas]

Hi!

The answer to your question is no, it is not possible.

A task (Linux process) is running on a single node only and can allocate/use only the available CPUs on that node. You need an MPI job (or Hybrid MPI+OpenMP) in order to utilize CPUs on different nodes for a single purpose/application.

Best Regards,
Chrysovalantis Paschoulas

On 09/09/2014 05:14 PM, Michal Zielinski wrote:
Phil,

I believe that 1 core per node is correct.

Maybe let me ask this first: is possible for a single task to use specific CPUs across several nodes?

Thanks,
Mike

On Tue, Sep 9, 2014 at 10:38 AM, Eckert, Phil <eckert2-i2BcT+NCU+***@public.gmane.org<mailto:eckert2-i2BcT+NCU+***@public.gmane.org>> wrote:
Mike,

In your slurm.conf you have Procs=1, (which is the same as CPUS=1) and Sockets (if ommited will be inferred from CPUS, default is 1) and CoresPerSocket (default is 1)

So at this point the slurm.conf has a default configuration of 1 core per node.

Phil Eckert
LLNL

From: Michal Zielinski <michal.zielinski-jZvHcCJSbXyHXe+***@public.gmane.org<mailto:***@uconn.edu>>
Reply-To: slurm-dev <slurm-dev-***@public.gmane.org<mailto:slurm-dev-***@public.gmane.org>>
Date: Tuesday, September 9, 2014 at 6:35 AM
To: slurm-dev <slurm-dev-***@public.gmane.org<mailto:slurm-dev-***@public.gmane.org>>
Subject: [slurm-dev] Re: "Requested node configuration is not available" when using -c

Josh,

I believe that -n sets the number of tasks. I only want a single task, as when a single process uses multiple cores. srun -n 2 hostname returns

linux-slurm2
linux-slurm3

which is definitely not what I want.

Thanks,
Mike


On Mon, Sep 8, 2014 at 8:07 PM, Josh McSavaney <McSaucy-s/YdZray+***@public.gmane.org<mailto:McSaucy-s/YdZray+***@public.gmane.org>> wrote:
I believe your slurm.conf is defining 4 nodes with a single logical processor each. You are then trying to allocate two CPUs on a single node with srun, which (according to your slurm.conf) you do not have.

You may want to consider `srun -n 2 hostname` and see where that lands you.

Regards,

Josh McSavaney
Bit Flipper
Rochester Institute of Technology



On Mon, Sep 8, 2014 at 7:42 PM, Christopher Samuel <samuel-***@public.gmane.org<mailto:samuel-***@public.gmane.org>> wrote:

On 09/09/14 07:26, Michal Zielinski wrote:

I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with
CR_CPU and no usage restrictions. Each node has just 1 CPU.
[...]
But, *srun -c 2 hostname* does not work, and it returns the above error.

I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?

cheers,
Chris
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
Email: samuel-***@public.gmane.org<mailto:samuel-***@public.gmane.org> Phone: [phone]<tel:%2B61%20%280%293%20903%2055545>
http://www.vlsci.org.au/ http://twitter.com/vlsci









------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------
Forschungszentrum Juelich GmbH
52425 Juelich
Sitz der Gesellschaft: Juelich
Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498
Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher
Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender),
Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt,
Prof. Dr. Sebastian M. Schmidt
------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------

[Michal Zielinski]

Thanks for your answer, it explains a lot.
I thought this was possible because it's possible using LSF without MPI,
but I guess not.
Thread closed?

Thanks,
Michal Zielinski

On Tue, Sep 9, 2014 at 11:22 AM, Chrysovalantis Paschoulas <

Post by Chrysovalantis Paschoulas
Hi!
The answer to your question is no, it is not possible.
A task (Linux process) is running on a single node only and can
allocate/use only the available CPUs on that node. You need an MPI job (or
Hybrid MPI+OpenMP) in order to utilize CPUs on different nodes for a single
purpose/application.
Best Regards,
Chrysovalantis Paschoulas
Phil,
I believe that 1 core per node is correct.
Maybe let me ask this first: is possible for a single task to use
specific CPUs across several nodes?
Thanks,
Mike

Post by Eckert, Phil
Mike,
In your slurm.conf you have Procs=1, (which is the same as CPUS=1) and
Sockets (if ommited will be inferred from CPUS, default is 1) and
CoresPerSocket (default is 1)
So at this point the slurm.conf has a default configuration of 1 core per node.
Phil Eckert
LLNL
Date: Tuesday, September 9, 2014 at 6:35 AM
Subject: [slurm-dev] Re: "Requested node configuration is not available" when using -c
Josh,
I believe that *-n *sets the number of tasks. I only want a single
task, as when a single process uses multiple cores. *srun -n 2 hostname*
returns
linux-slurm2
linux-slurm3
which is definitely not what I want.
Thanks,
Mike

Post by Josh McSavaney
I believe your slurm.conf is defining 4 nodes with a single logical
processor each. You are then trying to allocate two CPUs on a single node
with srun, which (according to your slurm.conf) you do not have.
You may want to consider `srun -n 2 hostname` and see where that lands you.
Regards,
Josh McSavaney
Bit Flipper
Rochester Institute of Technology
On Mon, Sep 8, 2014 at 7:42 PM, Christopher Samuel <

Post by Michal Zielinski
I have a small test cluster (node[1-4]) running slurm 14.03.0 setup

Post by Michal Zielinski
with
CR_CPU and no usage restrictions. Each node has just 1 CPU.

[...]

Post by Michal Zielinski
But, *srun -c 2 hostname* does not work, and it returns the above error.
I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?
cheers,
Chris
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
http://www.vlsci.org.au/ http://twitter.com/vlsci

------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------
Forschungszentrum Juelich GmbH
52425 Juelich
Sitz der Gesellschaft: Juelich
Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498
Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher
Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender),
Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt,
Prof. Dr. Sebastian M. Schmidt
------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------

[Carlos Fenoy]

Hi Michal,
What do you mean with "It's possible using LSF without MPI? as far as I
know a process can not use resources in different compute nodes. Only
distributed programming models would allow the usage of different compute
nodes but at least there must be a task per node, unless you use something
like ScaleMP, but that is a completely different matter.

Regards,
Carles Fenoy

Post by Michal Zielinski
Thanks for your answer, it explains a lot.
I thought this was possible because it's possible using LSF without MPI,
but I guess not.
Thread closed?
Thanks,
Michal Zielinski
On Tue, Sep 9, 2014 at 11:22 AM, Chrysovalantis Paschoulas <

Post by Chrysovalantis Paschoulas
Hi!
The answer to your question is no, it is not possible.
A task (Linux process) is running on a single node only and can
allocate/use only the available CPUs on that node. You need an MPI job (or
Hybrid MPI+OpenMP) in order to utilize CPUs on different nodes for a single
purpose/application.
Best Regards,
Chrysovalantis Paschoulas
Phil,
I believe that 1 core per node is correct.
Maybe let me ask this first: is possible for a single task to use
specific CPUs across several nodes?
Thanks,
Mike

Post by Eckert, Phil
Mike,
In your slurm.conf you have Procs=1, (which is the same as CPUS=1) and
Sockets (if ommited will be inferred from CPUS, default is 1) and
CoresPerSocket (default is 1)
So at this point the slurm.conf has a default configuration of 1 core per node.
Phil Eckert
LLNL
Date: Tuesday, September 9, 2014 at 6:35 AM
Subject: [slurm-dev] Re: "Requested node configuration is not available" when using -c
Josh,
I believe that *-n *sets the number of tasks. I only want a single
task, as when a single process uses multiple cores. *srun -n 2 hostname*
returns
linux-slurm2
linux-slurm3
which is definitely not what I want.
Thanks,
Mike

Post by Josh McSavaney
I believe your slurm.conf is defining 4 nodes with a single logical
processor each. You are then trying to allocate two CPUs on a single node
with srun, which (according to your slurm.conf) you do not have.
You may want to consider `srun -n 2 hostname` and see where that lands you.
Regards,
Josh McSavaney
Bit Flipper
Rochester Institute of Technology
On Mon, Sep 8, 2014 at 7:42 PM, Christopher Samuel <

Post by Michal Zielinski
I have a small test cluster (node[1-4]) running slurm 14.03.0 setup

Post by Michal Zielinski
with
CR_CPU and no usage restrictions. Each node has just 1 CPU.

[...]

Post by Michal Zielinski
But, *srun -c 2 hostname* does not work, and it returns the above error.
I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?
cheers,
Chris
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
http://www.vlsci.org.au/ http://twitter.com/vlsci

------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------
Forschungszentrum Juelich GmbH
52425 Juelich
Sitz der Gesellschaft: Juelich
Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498
Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher
Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender),
Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt,
Prof. Dr. Sebastian M. Schmidt
------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------

[Christopher Samuel]

Post by Michal Zielinski
I thought this was possible because it's possible using LSF without MPI,
but I guess not.

This is an operating system kernel issue, not a queuing system issue, so
Slurm, LSF or Torque will all have the same issue.

There are ways to build SMP systems out of individual nodes but they are
pretty rare. One company that does this is ScaleMP.

All the best,
Chris

[Christopher Samuel]

Post by Christopher Samuel
This is an operating system kernel issue, not a queuing system issue, so
Slurm, LSF or Torque will all have the same issue.

To clarify the Linux kernel can support these sorts of systems, but
you'll either need extra supporting software or specialised hardware,
or both. :-)

It's controlled by the kernel option CONFIG_X86_EXTENDED_PLATFORM, viz:

if X86_64
config X86_EXTENDED_PLATFORM
bool "Support for extended (non-PC) x86 platforms"
default y
---help---
If you disable this option then the kernel will only support
standard PC platforms. (which covers the vast majority of
systems out there.)

If you enable this option then you'll be able to select support
for the following (non-PC) 64 bit x86 platforms:
Numascale NumaChip
ScaleMP vSMP
SGI Ultraviolet

If you have one of these systems, or if you want to build a
generic distribution kernel, say Y here - otherwise say N.
endif

[Michal Zielinski]

To be honest I'm not sure about what setup was done (ex. MPI in the
background), but when I run *bsub -n 50 ./script* on our cluster, I get a
single process that has access to a total of 50 cores spread across
multiple nodes.Some of these nodes may also contain cores that are
allocated to a different job.

Post by Christopher Samuel

Post by Christopher Samuel
This is an operating system kernel issue, not a queuing system issue, so
Slurm, LSF or Torque will all have the same issue.

To clarify the Linux kernel can support these sorts of systems, but
you'll either need extra supporting software or specialised hardware,
or both. :-)
if X86_64
config X86_EXTENDED_PLATFORM
bool "Support for extended (non-PC) x86 platforms"
default y
---help---
If you disable this option then the kernel will only support
standard PC platforms. (which covers the vast majority of
systems out there.)
If you enable this option then you'll be able to select support
Numascale NumaChip
ScaleMP vSMP
SGI Ultraviolet
If you have one of these systems, or if you want to build a
generic distribution kernel, say Y here - otherwise say N.
endif
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
http://www.vlsci.org.au/ http://twitter.com/vlsci

[j***@public.gmane.org]

You have expressed this problem in a very misleading fashion. Here is
a table showing the equivalent commands and options for the most
common workload managers (Slurm, LSF, PBS, SGE and LoadLeveler) that
should help:
http://slurm.schedmd.com/rosetta.html

Post by Michal Zielinski
To be honest I'm not sure about what setup was done (ex. MPI in the
background), but when I run *bsub -n 50 ./script* on our cluster, I get a
single process that has access to a total of 50 cores spread across
multiple nodes.Some of these nodes may also contain cores that are
allocated to a different job.

Post by Christopher Samuel

Post by Christopher Samuel
This is an operating system kernel issue, not a queuing system issue, so
Slurm, LSF or Torque will all have the same issue.

To clarify the Linux kernel can support these sorts of systems, but
you'll either need extra supporting software or specialised hardware,
or both. :-)
if X86_64
config X86_EXTENDED_PLATFORM
bool "Support for extended (non-PC) x86 platforms"
default y
---help---
If you disable this option then the kernel will only support
standard PC platforms. (which covers the vast majority of
systems out there.)
If you enable this option then you'll be able to select support
Numascale NumaChip
ScaleMP vSMP
SGI Ultraviolet
If you have one of these systems, or if you want to build a
generic distribution kernel, say Y here - otherwise say N.
endif
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
http://www.vlsci.org.au/ http://twitter.com/vlsci

[Josh McSavaney]

Hey Mike,

I had assumed that this is what you were trying to accomplish. Seeing as
your are trying to create a single task with multiple cores, you must first
define a node that has enough cores.

As far as SLURM knows, you currently have 1 CPU core per node (as Phil
stated). SLURM is refusing to over-allocate CPUs. Try increasing your CPU
count beyond 1 for a node and SLURM should be happy with allocating more
than 1 CPU core.

I've attached a file to show a slice of my config along with a few srun
submissions. Ignore the `--qos free` portions of the srun statements as
these are required for my configuration.

Hope this helps,

Josh McSavaney
Bit Flipper
Rochester Institute of Technology

Post by Michal Zielinski
Josh,
I believe that *-n *sets the number of tasks. I only want a single task,
as when a single process uses multiple cores. *srun -n 2 hostname* returns
linux-slurm2
linux-slurm3
which is definitely not what I want.
Thanks,
Mike

Post by Josh McSavaney
I believe your slurm.conf is defining 4 nodes with a single logical
processor each. You are then trying to allocate two CPUs on a single node
with srun, which (according to your slurm.conf) you do not have.
You may want to consider `srun -n 2 hostname` and see where that lands you.
Regards,
Josh McSavaney
Bit Flipper
Rochester Institute of Technology

Post by Michal Zielinski
I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with

Post by Michal Zielinski
CR_CPU and no usage restrictions. Each node has just 1 CPU.

[...]

Post by Michal Zielinski
But, *srun -c 2 hostname* does not work, and it returns the above error.
I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?
cheers,
Chris
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
http://www.vlsci.org.au/ http://twitter.com/vlsci

[Martin Perry]

Michal,
The job has only 1 task. -c is cpus per task. Each task can run on only one node. So if each node has only 1 cpu, you can only assign 1 cpu per task.
Martin

From: Michal Zielinski [mailto:***@uconn.edu]
Sent: Tuesday, September 09, 2014 6:36 AM
To: slurm-dev
Subject: [slurm-dev] Re: "Requested node configuration is not available" when using -c

Josh,
I believe that -n sets the number of tasks. I only want a single task, as when a single process uses multiple cores. srun -n 2 hostname returns
linux-slurm2
linux-slurm3
which is definitely not what I want.

Thanks,
Mike


On Mon, Sep 8, 2014 at 8:07 PM, Josh McSavaney <***@csh.rit.edu<mailto:***@csh.rit.edu>> wrote:
I believe your slurm.conf is defining 4 nodes with a single logical processor each. You are then trying to allocate two CPUs on a single node with srun, which (according to your slurm.conf) you do not have.

You may want to consider `srun -n 2 hostname` and see where that lands you.

Regards,

Josh McSavaney
Bit Flipper
Rochester Institute of Technology


On Mon, Sep 8, 2014 at 7:42 PM, Christopher Samuel <***@unimelb.edu.au<mailto:***@unimelb.edu.au>> wrote:

On 09/09/14 07:26, Michal Zielinski wrote:
I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with
CR_CPU and no usage restrictions. Each node has just 1 CPU.
[...]
But, *srun -c 2 hostname* does not work, and it returns the above error.

I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?

cheers,
Chris

[Michal Zielinski]

Attached is the output.

Thanks,
Mike

Post by Michal Zielinski
I have a small test cluster (node[1-4]) running slurm 14.03.0 setup with

Post by Michal Zielinski
CR_CPU and no usage restrictions. Each node has just 1 CPU.

[...]

Post by Michal Zielinski
But, *srun -c 2 hostname* does not work, and it returns the above error.
I have no idea why I can't dedicate 2 cores to a single job if I can
dedicate each core individually to a job.

What does "scontrol show node" say?
cheers,
Chris
--
Christopher Samuel Senior Systems Administrator
VLSCI - Victorian Life Sciences Computation Initiative
http://www.vlsci.org.au/ http://twitter.com/vlsci