2. Determine the file system name. On the primary MGMT node:
[admin@sn11000n000 ~]$
cscli fs_info
Examples show a file system named
sn11000
:
[admin@sn11000n000 ~]$
cscli fs_info
----------------------------------------------------------------------------------------------------
OST Redundancy style: Declustered Parity (GridRAID)
Disk I/O Integrity guard (ANSI T10-PI) is not supported by hardware
----------------------------------------------------------------------------------------------------
Information about "sn11000" file system:
----------------------------------------------------------------------------------------------------
Node Role Targets Failover partner Devices
----------------------------------------------------------------------------------------------------
sn11000n000 mgmt 1 / 0 sn11000n001
sn11000n001 mgmt 1 / 0 sn11000n000
sn11000n002 mgs 0 / 1 sn11000n003 /dev/md65
sn11000n003 mds 0 / 1 sn11000n002 /dev/md66
sn11000n004 oss 0 / 1 sn11000n005 /dev/md0
sn11000n005 oss 0 / 1 sn11000n004 /dev/md1
3. If the output in the previous command indicates that the Lustre file system is not started (e.g., the
Targets
column shows
0 / 1
instead of
1 / 1
for OSS nodes), start the Lustre file system:
[admin@sn11000n000 ~]$
cscli mount -f testfs
mount: MGS is starting...
mount: MGS is started!
mount: testfs is started on sn11000n002,sn11000n003!
mount: testfs is started on sn11000n004,sn11000n005!
mount: All start commands for filesystem testfs were sent.
mount: Use "cscli show_nodes" to see mount status.
[admin@sn11000n000 ~]$
cscli show_nodes
--------------------------------------------------------------------------------
---------
Hostname Role Power State Service State Targets HA Partner HA
Resources
--------------------------------------------------------------------------------
---------
sn11000n000 MGMT On Started 1 / 0 sn11000n001 Local
sn11000n001 MGMT On Started 1 / 0 sn11000n000 Local
sn11000n002 MDS,MGS On Started 1 / 1 sn11000n003 Local
sn11000n003 MDS,MGS On Started 1 / 1 sn11000n002 Local
sn11000n004 OSS On Started 1 / 1 sn11000n005 Local
sn11000n005 OSS On Started 1 / 1 sn11000n004 Local
--------------------------------------------------------------------------------
---------
4. Log in to the secondary MGMT node:
[MGMT0]$
ssh -l admin secondary_MGMT_node
5. Change to root user on the secondary MGMT node:
[MGMT1]$
sudo su -
6. Obtain the Lustre network identifier (NIDs) of each MGS node (typically n002 and n003:
[MGMT1]#
ssh n002 lctl list_nids
[MGMT1]#
ssh n003 lctl list_nids
Perform the First-Run Configuration
88
