Smhp add features in LCS utills #392
Conversation
| smhp_conf="/opt/ml/config/resource_config.json" | ||
|
|
||
| pecho(){ | ||
| #pretty echo |
| # node_type="$(cat "${smhp_conf}" | jq -r ".InstanceGroups[] | select(.Instances[0].CustomerIpAddress == \"${hip}\") | .Name")" | ||
| # [[ -n $node_type ]] && break | ||
| # done | ||
| pecho "node_type=${node_type}" # controller-machine, worker-group-1 |
There was a problem hiding this comment.
Perhaps in the comment add Ex, it reads more naturally # Ex: controller-machine, worker-group-1
|
|
||
| if [[ $node_type == "controller-machine" ]] ;then | ||
| slurm_fix | ||
| # scontrol reconfigure |
| @@ -0,0 +1,75 @@ | |||
| #!/bin/bash | |||
| # https://slurm.schedmd.com/pam_slurm_adopt.html | |||
There was a problem hiding this comment.
Add a short sentence at the beginning to say what it does, people will not click on every link to understand what it does
| @@ -0,0 +1,88 @@ | |||
| #!/bin/bash | |||
There was a problem hiding this comment.
No generation of ssh keys for root.
| @@ -0,0 +1,88 @@ | |||
| #!/bin/bash | |||
| # https://slurm.schedmd.com/pam_slurm_adopt.html | |||
There was a problem hiding this comment.
Same references as in the slurm_fix. Remove or can you explain why use use them here too. What does this file do too? Add a small note on the header, 1 sentence is ok, more if needed.
|
|
||
| if ! [[ -f "${ssh_dir}/${key_priv}" ]] ;then | ||
| rm -f "${ssh_dir}/${key_pub}" | ||
| # rm -f "${ssh_dir}/${key_auth}" |
| @@ -0,0 +1,249 @@ | |||
| #!/bin/bash | |||
| # https://slurm.schedmd.com/pam_slurm_adopt.html | |||
There was a problem hiding this comment.
copy paste from other files?
There was a problem hiding this comment.
No idea what the files does too, add a header
| pyxis_conf="${plugstack_dir}/pyxis.conf" | ||
|
|
||
| smhp_conf="/opt/ml/config/resource_config.json" | ||
| # apt_opts='-o Dpkg::Options::="--force-confdef" -o Dpkg::Options::="--force-confold"' |
| scancel -u ubuntu | ||
| squeue | ||
|
|
||
| # ssh ip-10-1-64-87 uname -a ; -u ubuntu ssh ip-10-1-64-87 uname |
| squeue | ||
| ssh $node uname | ||
| # Access denied by pam_slurm_adopt: you have no active jobs on this node | ||
| # Connection closed by 10.1.64.87 port 22 |
|
|
||
|
|
||
| # test | ||
| node="ip-10-1-64-87" |
| grep -Hn "^UsePAM " ${sshd_conf} | ||
|
|
||
| # apt update | ||
| # # -o Dpkg::Options::="--force-confdef" -o Dpkg::Options::="--force-confold" |
| @@ -0,0 +1,75 @@ | |||
| #!/bin/bash | |||
There was a problem hiding this comment.
Please rename this script to an explicit functionality.
This is unclear.
|
Any update on this @gmgtamz ? |
3 scripts
ssh_root.sh to add the SSH root capability
slurm_fix.sh to fix the plugstack bug
pam_adopt_cgroup_wheel.sh to add cgroup, adoptpam and wheel groups
none of them is linked into the python caller script (lifecycle_script.py) yet