General procedure for powering off an external ClusterStor system.
Use this procedure as a general guide to power off an external ClusterStor system. Refer to the detailed procedures in the appropriate ClusterStor administration guide:
Title | Model |
---|---|
ClusterStor E1000 Administration Guide 4.2 - S-2758 |
ClusterStor E1000 |
ClusterStor Administration Guide 3.4 - S-2756 |
ClusterStor L300/L300N |
ClusterStor Administration Guide - S-2755 |
Legacy ClusterStor |
(remote#
) SSH to the primary MGMT node as admin
.
ssh -l admin cls01234n00.us.cray.com
(n000$
) Change to root
user.
sudo su –
(n000#
) Collect status information for the system before shutdown.
cscli csinfo
cscli show_nodes
cscli fs_info
crm_mon -1r
(n000#
) Check resources before unmounting the file system.
ssh cls01234n002 crm_mon -r1 | grep fsys
ssh cls01234n004 crm_mon -r1 | grep fsys
ssh cls01234n006 crm_mon -r1 | grep fsys
ssh cls01234n008 crm_mon -r1 | grep fsys
ssh cls01234n010 crm_mon -r1 | grep fsys
ssh cls01234n012 crm_mon -r1 | grep fsys
. . .
(n000#
) Stop the Lustre file system (FILESYSTEM_NAME
will be reported from the cscli fs_info
command run above).
cscli unmount -f FILESYSTEM_NAME
(n000#
) Verify that resources have been stopped by running the following on all even-numbered nodes.
ssh NODENAME crm_mon -r1 | grep fsys
Example output:
cls01234n006_md0-fsys (ocf::heartbeat:XYMNTR): Stopped
cls01234n006_md1-fsys (ocf::heartbeat:XYMNTR): Stopped
cls01234n006_md2-fsys (ocf::heartbeat:XYMNTR): Stopped
cls01234n006_md3-fsys (ocf::heartbeat:XYMNTR): Stopped
cls01234n006_md4-fsys (ocf::heartbeat:XYMNTR): Stopped
cls01234n006_md5-fsys (ocf::heartbeat:XYMNTR): Stopped
cls01234n006_md6-fsys (ocf::heartbeat:XYMNTR): Stopped
cls01234n006_md7-fsys (ocf::heartbeat:XYMNTR): Stopped
(n000#
) SSH to the MGS node (the MGS_NODE
name will be reported from the cscli fs_info
command run above).
ssh MGS_NODE
(mgs#
) To determine if Resource Group md65-group
is stopped, use the crm_mon
utility to monitor the status of the MGS and MDS nodes.
Shows MGS and MDS nodes in a partial stopped state.
[MGS]# crm_mon -1r | grep fsys
Example output:
cls01234n003_md66-fsys (ocf::heartbeat:XYMNTR): Stopped
cls01234n003_md65-fsys (ocf::heartbeat:XYMNTR): Started
If the output of the previous command shows a partial stopped state (Stopped
and Started
), issue the stop_xyraid
command and verify that the node is stopped.
[MGS]# stop_xyraid nodename_md65-group
[MGS]# crm\_mon -1r | grep fsys
Example output:
cls01234n003_md66-fsys (ocf::heartbeat:XYMNTR): Stopped
cls01234n003_md65-fsys (ocf::heartbeat:XYMNTR): Stopped
(mgs#
) Exit the MGS node.
exit
(n000#
) Power off the non-MGMT diskless nodes.
Check power state of all non-MGMT nodes and list the node hostnames (in this example cls01234n[02-15]
) before power off.
pm -q
Example output:
on: cls01234n[000-001]
on: cls01234n[002-015]
unknown:
Power off all non-MGMT nodes.
cscli power_manage -n cls01234n[02-15] --power-off
Check the power status of the nodes.
pm -q
Example output:
on: cls01234n[000-001]
off: cls01234n[002-015]
unknown:
(n000#
) Repeat the previous step until all non-MGMT nodes are powered off.
(n000#
) From the primary MGMT node, power off the MGMT nodes.
cscli power_manage -n cls01234n[000-001] --power-off
(n000#
) Shut down the primary management node.
shutdown -h now
Return to System Power Off Procedures and continue with next step.