HPC: use script_run for munge keys and slurm conf

Both, munge and slurm configs are scp using custom util method: exec_and_insert_password. The patch is trying to simplify the approach and thus make it more reliable by making sure the ssh keys are distributed to all nodes so that direct usage of scp is allowed.
os-autoinst · Dec 4, 2019 · c18cca9 · c18cca9
1 parent 19308ef
commit c18cca9
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 6 deletions.
diff --git a/lib/hpc/configs.pm b/lib/hpc/configs.pm
@@ -23,14 +23,14 @@ sub prepare_slurm_conf {
 
     if ($slurm_conf eq "basic") {
         my $config = << "EOF";
-sed -i "/^ControlMachine.*/c\\ControlMachine=$cluster_ctl_nodes[0]" /etc/slurm/slurm.conf
+sed -i "/^SlurmctldHost.*/c\\SlurmctldHost=$cluster_ctl_nodes[0]" /etc/slurm/slurm.conf
 sed -i "/^NodeName.*/c\\NodeName=$cluster_ctl_nodes,$cluster_compute_nodes Sockets=1 CoresPerSocket=1 ThreadsPerCore=1 State=unknown" /etc/slurm/slurm.conf
 sed -i "/^PartitionName.*/c\\PartitionName=normal Nodes=$cluster_ctl_nodes,$cluster_compute_nodes Default=YES MaxTime=24:00:00 State=UP" /etc/slurm/slurm.conf
 EOF
         assert_script_run($_) foreach (split /\n/, $config);
     } elsif ($slurm_conf eq "accounting") {
         my $config = << "EOF";
-sed -i "/^ControlMachine.*/c\\ControlMachine=$cluster_ctl_nodes[0]" /etc/slurm/slurm.conf
+sed -i "/^SlurmctldHost.*/c\\SlurmctldHost=$cluster_ctl_nodes[0]" /etc/slurm/slurm.conf
 sed -i "/^NodeName.*/c\\NodeName=$cluster_ctl_nodes,$cluster_compute_nodes Sockets=1 CoresPerSocket=1 ThreadsPerCore=1 State=unknown" /etc/slurm/slurm.conf
 sed -i "/^PartitionName.*/c\\PartitionName=normal Nodes=$cluster_ctl_nodes,$cluster_compute_nodes Default=YES MaxTime=24:00:00 State=UP" /etc/slurm/slurm.conf
 sed -i "/^#JobAcctGatherType.*/c\\JobAcctGatherType=jobacct_gather/linux" /etc/slurm/slurm.conf
@@ -42,7 +42,7 @@ EOF
         assert_script_run($_) foreach (split /\n/, $config);
     } elsif ($slurm_conf eq "ha") {
         my $config = << "EOF";
-sed -i "/^ControlMachine.*/c\\ControlMachine=$cluster_ctl_nodes[0]" /etc/slurm/slurm.conf
+sed -i "/^SlurmctldHost.*/c\\SlurmctldHost=$cluster_ctl_nodes[0]" /etc/slurm/slurm.conf
 sed -i "/^#BackupController.*/c\\BackupController=$cluster_ctl_nodes[1]" /etc/slurm/slurm.conf
 sed -i "/^StateSaveLocation.*/c\\StateSaveLocation=/shared/slurm/" /etc/slurm/slurm.conf
 sed -i "/^NodeName.*/c\\NodeName=$cluster_ctl_nodes,$cluster_compute_nodes Sockets=1 CoresPerSocket=1 ThreadsPerCore=1 State=unknown" /etc/slurm/slurm.conf
@@ -53,7 +53,7 @@ EOF
         assert_script_run($_) foreach (split /\n/, $config);
     } elsif ($slurm_conf eq "nfs_db") {
         my $config = << "EOF";
-sed -i "/^ControlMachine.*/c\\ControlMachine=$cluster_ctl_nodes[0]" /etc/slurm/slurm.conf
+sed -i "/^SlurmctldHost.*/c\\SlurmctldHost=$cluster_ctl_nodes[0]" /etc/slurm/slurm.conf
 sed -i "/^#BackupController.*/c\\BackupController=$cluster_ctl_nodes[1]" /etc/slurm/slurm.conf
 sed -i "/^StateSaveLocation.*/c\\StateSaveLocation=/shared/slurm/" /etc/slurm/slurm.conf
 sed -i "/^NodeName.*/c\\NodeName=$cluster_ctl_nodes,$cluster_compute_nodes Sockets=1 CoresPerSocket=1 ThreadsPerCore=1 State=unknown" /etc/slurm/slurm.conf

diff --git a/lib/hpcbase.pm b/lib/hpcbase.pm
@@ -85,15 +85,15 @@ sub distribute_munge_key {
     my ($self) = @_;
     my @cluster_nodes = cluster_names();
     foreach (@cluster_nodes) {
-        exec_and_insert_password("scp -o StrictHostKeyChecking=no /etc/munge/munge.key root\@$_:/etc/munge/munge.key");
+        script_run("scp -o StrictHostKeyChecking=no /etc/munge/munge.key root\@$_:/etc/munge/munge.key");
     }
 }
 
 sub distribute_slurm_conf {
     my ($self) = @_;
     my @cluster_nodes = cluster_names();
     foreach (@cluster_nodes) {
-        exec_and_insert_password("scp -o StrictHostKeyChecking=no /etc/slurm/slurm.conf root\@$_:/etc/slurm/slurm.conf");
+        script_run("scp -o StrictHostKeyChecking=no /etc/slurm/slurm.conf root\@$_:/etc/slurm/slurm.conf");
     }
 }
 

diff --git a/tests/hpc/slurm_master.pm b/tests/hpc/slurm_master.pm
@@ -219,6 +219,7 @@ sub run {
     my $nodes      = get_required_var('CLUSTER_NODES');
     my $slurm_conf = get_required_var('SLURM_CONF');
     $self->prepare_user_and_group();
+    $self->generate_and_distribute_ssh();
 
     # provision HPC cluster, so the proper rpms are installed,
     # munge key is distributed to all nodes, so is slurm.conf