PBS PRO Scheduler Configuration:

PBS PRO Scheduler application Configuration:

How to add new application on nodes

Below is the command to change the existing configuration, see here on node1.example.com (compute node1) only two applications are allowed to run jobs that is Abaqus & Dyna_MPP.

Now the requirement is user wants to run Pamcrash application on node1.autoliv.int, For enabling that please follow the below steps

BEFORE ADDING APPLICATION

#pbsnode -a

node1.example.com

     Mom = node1.example.com

     Port = 15002

     pbs_version = PBSPro_13.1.0.160576

     ntype = PBS

     state = free

     pcpus = 32

     resources_available.arch = linux

     resources_available.host = node1

     resources_available.mem = 132279240kb

     resources_available.ncpus = 32

     resources_available.pas_applications_enabled = Abaqus,Dyna_Mpp

     resources_available.vnode = node1.example.com

     resources_assigned.accelerator_memory = 0kb

     resources_assigned.mem = 0kb

     resources_assigned.naccelerators = 0

     resources_assigned.ncpus = 0

     resources_assigned.netwins = 0

     resources_assigned.ngpus = 0

     resources_assigned.vmem = 0kb

     resv_enable = True

     sharing = default_shared

ADDING NEW APPLICATION ON NODE (node1.example.com is a hostname of that node)

 # qmgr

Max open servers: 49

Qmgr: set node node1.example.com resources_available.pas_applications_enabled+=Pamcrash

AFTER ADDING APPLICATION

#pbsnode –a

node1.example.com

     Mom = node1.example.com

     Port = 15002

     pbs_version = PBSPro_13.1.0.160576

     ntype = PBS

     state = free

     pcpus = 32

     resources_available.arch = linux

     resources_available.host = node1

     resources_available.mem = 132279240kb

     resources_available.ncpus = 32

     resources_available.pas_applications_enabled = Abaqus,Dyna_Mpp,Pamcrash

     resources_available.vnode = node1.example.com

     resources_assigned.accelerator_memory = 0kb

     resources_assigned.mem = 0kb

     resources_assigned.naccelerators = 0

     resources_assigned.ncpus = 0

     resources_assigned.netwins = 0

     resources_assigned.ngpus = 0

     resources_assigned.vmem = 0kb

     resv_enable = True

     sharing = default_shared

-bash: fork: Cannot allocate memory in RHEL 6/CentOS 6

-bash: fork: Cannot allocate memory in Red Hat Enterprise Linux 6 (or) CentOS 6

Solution:-

This problem is caused because of Process Identifiers Limit, Increasing the value will help on large Linux system or clusters to ease process identification and process management, So that we can avoid this kind error “-bash: fork: Cannot allocate memory”

If you want to check current value execute this command

# cat /proc/sys/kernel/pid_max

32768

(OR)

# sysctl kernel.pid_max

kernel.pid_max = 32768

So I am going to increase this value to 65534.

# echo kernel.pid_max = 65534 >> /etc/sysctl.conf

To configure kernel parameters at runtime use sysctl command

# sysctl –p

{Note: p option is used to load sysctl setting from the file      /etc/sysctl.conf }