HP XC System Software Installation Guide Version 2.1

18. Start the cluster_fstab service t o merge the changes fro m the fstab.proto file
into the /etc/fstab file on the head node:
# service cluster_fstab start
19. Ensure that NIS logins are accepted, and recycle the sshd service on the head node:
# service sshd restart
20. When the head nod e i s con figured as a slave NIS server, you may notice that the head node
does not accept NIS logins after it has been rebo oted. If this happens, check the status of
the indicated services by issuing the following comm and s. Look for error messages that
indicateiftheservicedidnotstart properly. For example:
# service ypserv status
ypserv dead but pid file exists
# service ypxfrd status
ypxfrd is stopped
# service ypbind status
ypbind is stopped
If any of the services are not running (that is, they are stopped or dead), restart all of these
services in the following o rd e r:
# service ypserv restart
# service ypxfrd restart
# service ypbind restart
Yo u mu st periodically synchro nize the NIS database on the head node with the database on the
master NIS server. You can accomplish this by creating a cron job to automate periodic
updates. The HP XC System Software Administration Guide provides a procedure that d escribes
how to define the cron job. You can also refer to ypxfr
(8) for more information.
4.7.5 Limit User Access to Compute N odes (Optional)
By default, all compute nodes in the XC system are accessible by any user after their user
accounts have b een set up. By enabling the pluggable authentication module (PAM),
pam_slurm, you can restrict individual access to each compute node to the user who currently
has the compute node reserved within SLURM.
________________________ Caution _______________________
The procedure documented here establishes restricted access on all nodes with the
compute role. By default, all nodes are assigned the compute role.
When you run th e cluster_config command described in Section 4.8, it is
important that you assign a login role to each node on which you expect users to
be ab le to log in and use the system.
If you enable the PAM module here and configure all nodes with the com pu te role
(the default), but you do no t configure any nodes with the l ogin role, no n-roo t users
will not be allowed to log in to the system.
Follow this procedure to limit user access to compute nodes:
1. Use the text editor of y
our choice to edit the foll owi ng file:
/opt/hptc/slurm/etc/nconfig.d/munge_nconfig.pl
2. Change the following li ne from th is:
my $ENABLE_PAM_SLURM = 0;
to this:
my $ENABLE_PAM_SLURM = 1;
Configuring and Imaging the System 4-13