Hey Sam,
as you can see, no cgroup_disable anywhere:
[root@frodo ~]# cat /proc/cmdline
init=/sbin/bootcpuset root=/dev/md0 rd.luks=0 rd.lvm=0 LANG=en_US.UTF-8
rd_MD_UUID=6fdc973b:49b47f48:e87de278:ae21f48c SYSFONT=latarcyrheb-sun16
crashkernel=512M KEYBOARDTYPE=pc KEYTABLE=us rd.dm=0
ipmi_si.trydefaults=0 ipmi_si.trydmi=0 rcutree.rcu_cpu_stall_suppress=1
add_efi_memmap nortsched processor.max_cstate=1 intel_idle.max_cstate=1
nmi_watchdog=0 ipmi_si.tryacpi=0 nobau log_buf_len=8M
pci=hpiosize=0,hpmemsize=0,nobar earlyprintk=ttyS0,115200n8
relax_domain_level=2 cgroup_disable=memory nohz=off pcie_aspm=on
highres=off rd.shell udev.children-max=64 console=ttyS0,115200n8
But thanks for your help so far :)
Adam
Am 19.07.2016 um 19:15 schrieb Sam Gallop (NBI):
Hi Adam,
Sorry for not replying sooner. What does the output of 'cat /proc/cmdline'
show? Does it have cgroup_disable in there anywhere?
---
Sam Gallop
-----Original Message-----
From: A. Podstawka [mailto:adam.podsta...@dsmz.de]
Sent: 13 July 2016 12:24
To: slurm-dev <slurm-dev@schedmd.com>
Subject: [slurm-dev] RE: SGI UV2000 with SLURM
Hi Samuel,
[root@frodo ~]# slurmd -C
ClusterName=(null) NodeName=frodo CPUs=128 Boards=8 SocketsPerBoard=16
CoresPerSocket=8 ThreadsPerCore=1 RealMemory=1998969 TmpDisk=273505
UpTime=19-01:49:04
this seems equivalent to the "configline".
We are running RHEL7 with the /sbin/bootcpuset but i think i need to use the
default init system from rhel7 and cgroups instead of the sgi own bootcpuset
binary.
My problem is, with the bootcpcuset active, slurm can only start/run jobs
inside of the (very small: cpu=0-7) bootcpuset.
any good advices or hints, links to a good cgroups manual?
Thanks
Adam
Am 08.07.2016 um 12:51 schrieb Sam Gallop (NBI):
Hi Adam,
What does 'slurmd -C' say? If you don't have SLURM installed yet running the
SGI command 'topology' or 'lscpu' may give you an insight of what should be
specified.
If you are having trouble with cgroups have you checked that it's not been
disabled as a kernel parameter?
---
Samuel Gallop
Computing infrastructure for Science
CiS Support & Development
NBI Partnership Ltd.
Norwich Research Park
Colney Lane, Norwich
NR4 7UH
The NBI Partnership Ltd provides non-scientific services to the
Earlham Institute, the Institute of Food Research, the John Innes
Centre and The Sainsbury Laboratory
-----Original Message-----
From: A. Podstawka [mailto:adam.podsta...@dsmz.de]
Sent: 08 July 2016 08:05
To: slurm-dev <slurm-dev@schedmd.com>
Subject: [slurm-dev] SGI UV2000 with SLURM
Hi,
just a small question, is here anyone using an SGI UV2000 with SLURM?
We want to migrate from SGE to SLURM, but have some trouble to get slurm with
cgroups (cpusets) correctly running on our SGI UV2000.
Our Nodeline:
NodeName=frodo NodeAddr=172.18.250.11 CPUs=128 RealMemory=1998969
CoresPerSocket=8 ThreadsPerCore=1 State=DOWN
is this correct for an SGI UV with 8 "Nodes" with 2 sockets per node
and
8 cores per socket? As this is a aggregated system, i have a hard time to set
it right.
Thanks in advance
Adam
--
Adam Podstawka
Senior Systemadministrator
Informatician - Bioinformatics
Leibniz Institute DSMZ-German Collection of Microorganisms and Cell
Cultures Inhoffenstr. 7 B
38124 Braunschweig
Germany
www.dsmz.de
Director: Prof. Dr. Jörg Overmann
Local court: Braunschweig HRB 2570
Chairman of the management board: RD Dr. David Schnieders
DSMZ - A member of the Leibniz Association www.leibniz-gemeinschaft.de
--
Adam Podstawka
Senior Systemadministrator
Informatician - Bioinformatics
Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures
Inhoffenstr. 7 B
38124 Braunschweig
Germany
www.dsmz.de
Director: Prof. Dr. Jörg Overmann
Local court: Braunschweig HRB 2570
Chairman of the management board: RD Dr. David Schnieders
DSMZ - A member of the Leibniz Association www.leibniz-gemeinschaft.de
