HPE Performance Cluster Manager command reference
Power commands
- General format for power commands
cm power action -t target hostname
- Obtain power status for a node
cm power status -t node node_hostname
- Power on the cluster
cm power on -t system
- Power off the cluster
cm power off -t system
- Identify node
cm power identify -t node node_hostname -i 60
Display node hostnames
- Display hostnames for all nodes
cnodes
- Display leader node hostnames
cnodes --leader
- Display non-ICE compute node hostnames
cnodes --compute
- Display ICE compute node hostnames
cnodes --ice-compute
- Display chassis manager controller (CMC) hostnames
cnodes --cmc
- Display non-ICE compute nodes that are offline
cnodes --offline
Multinode queries
- Query non-ICE compute nodes
pdsh -g compute query
- Query ICE compute nodes
pdsh -g ice-compute query
- Query leader nodes
pdsh -g leader query
- Display the number of non-ICE compute nodes that currently report a certain speed
pdsh -g compute ethtool device_name | grep -c "speed"
For example:pdsh -g compute ethtool eth0 | grep -c "1000mb/s"
- Display NIC speed outliers
pdsh -g compute ethtool device_name | grep Speed | dshbak -c
For example:
pdsh -g compute ethtool eth0 | grep Speed | dshbak -c
- Display non-ICE compute nodes with unexpected memory size
pdsh -g compute grep MemTotal /proc/meminfo | dshbak -c
- Display non-ICE compute nodes with unexpected load averages
pdsh -g compute uptime
- Display the boot file systems that are mounted
pdsh -g compute df -h | grep -c boot
- Increase/decrease number of simultaneous commands
pdsh -f number group query
Repository management commands
- Display all repositories
cm repo show
- Display selected repositories
cm repo show | grep ^\*
- Display unselected repositories
cm repo show | grep -v ^\*
- Make repository available to provisioning commands
cm repo select repository_name
- Make repository unavailable to provisioning commands
cm repo unselect repository_name
- Refresh custom repository metadata
cm repo refresh repository_name
- Add custom repository
cm repo add /opt/clmgr/repos/mypkgs --custom mypkgs
Configure the cluster
- Add nodes
discover --node number
- Display path to cluster definition file
discover --show-configfile
Manage images and provision
- Copy files into an image
cp /etc/myfile /opt/clmgr/image/images/image_name/etc
- Display managed images and image kernel versions
cm image show
- Display image and kernel to be installed at next image operation
cinstallman --show-nodes
- From a flat compute node, display the image that was used on boot
cat /proc/cmdline
- From the admin node, display the image that was used to boot all non-ICE compute nodes
pdsh -g compute cat /proc/cmdline
- Assign an image and kernel version to a node for future imaging
cinstallman --assign-image --node hostname --image image_name --kernel version
- Reimage a node the next time it boots
cinstallman --next-boot image --node hostname
- Specify that a node boot from its Ethernet device the next time it boots
ipmiwrapper hostname chassis bootdev pxe
- Create an image from selected repositories and an RPM list
cm image create -i new_image -l /opt/clmgr/image/rpmlists/my-rhel7.6.rpmlist
- Create a new image from an existing image
cm image copy -s existing_image -i new_image
- Capture an image from a running non-ICE compute node
cm image capture --image image_name --node number
- Capture an image from a running non-ICE compute node and exclude some structures
cm image capture --image image_name --node hostname --exclude /var/spool/BBS/\\*_logs/\\* --exclude /var/spool/PBS/server_priv/accounting/\\*
- Installing RPMs into an image
RHEL:
cm image yum --duk -i image_name install RPM RPM ...
SLES:
cm image zypper --duk -i image_name install RPM RPM ...
- Removing RPMs from an image
RHEL:
cm image yum --duk -i image_name remove RPM RPM ...
SLES:
cm image zypper --duk -i image_name remove RPM RPM ...
- Installing RLMs into an image on a running non-ICE compute node
RHEL:
cm node yum -n hostname install RPM RPM ...
SLES:
cm node zypper -n hostname install RPM RPM ...
- Removing RPMs from an image on a running non-ICE compute node
RHEL:
cm image yum --duk -n hostname remove RPM RPM ...
SLES:
cm image zypper --duk -n hostname remove RPM RPM ...
Commands for ICE compute nodes
- Pushing ICE compute node images to a rack
cimage --push-rack image_name rack_number
For rack_number, specify a rack number in the
rX
format.- Display ICE compute node images and kernels
cimage --show-images
- Display image, kernel, and file system mode of ICE compute nodes
cimage --show-nodes rack_number
For rack_number, specify a rack number in the
rX
format.- Change the image, kernel, or file system mode of ICE compute nodes
cimage --set [option] image_name kernel node
Update RPMs
- Update RPMs on a node
cinstallman --update-node --node hostname
- Update RPMs on an image
cinstallman --update-image --image image_name
VCS commands
- Check in an image with changes or updates into version control
cm image revision commit -i image_name -m "checking in"
- Review version control history for an image
cm image revision history -i image_name
- Revert to a previous image version
cm image revision revert -i image_name --rev version_number
Directories
- Review post-installation scripts
ls /opt/clmgr/image/scripts/post-install
- Configuration scripts directory
/opt/clmgr/image/images/image_name/etc/opt/sgi/conf.d
- Set up one-time PXE boot for HPE ProLiant DL360 Gen10, HPE ProLiant DL380 Gen10, and HPE ProLiant DL380 Gen9 server installation
Use one of the following methods to configure a one-time PXE boot on nodes with iLO devices:
From the admin node, use the following
ilorest
command:ilorest bootorder --onetimeboot=pxe --url=iLO_IP -u ADMIN -p iLO_password --commit
For iLO_IP, ADMIN, and iLO_password, enter the values for this cluster.
Use
pdsh
orssh
to log into a node, and then run one of the following commands:hpbootcfg -P -b
or
efibootmgr options ...
On the node console, press F12 or Esc+@ (serial).
You can download
ilorest
andhpbootcfg
from the HPE support website.