cboot - program to power and or boot Cplant(tm) cluster
base
Cboot is a front end program to be used for management tasks such as
powering and booting of the Cplant(tm) Cluster system.
cboot [OPTION]
Where n is node-number
Where R is Rack-number
Where < text > is the specification required by the OPTION switch
Where [ text ] is the specification that is allowed but not required by the OPTION switch.
action(s) on the specified collection(s)
or device(s) to boot only.
action(s) on the specified collection(s)
or device(s) to power "off", "on" or "cycle", "cycle" is the default and will
take place if no other action is specified.
Note: the absense of both the --power and --boot switches causes cboot to accomplish powering followed by booting the specified nodes.
--leader -t 0 will specify the required action
to be executed on the leader(s) of the nodes in the collection t-0.
cboot -t 0 -n 0 --libdebug Device::Node::Alpha
Rack Designations:
Color Physical Label device specification
Red (R0-Tn) r-n
Teal (T0-Tn) t-n
Green (G0-Gn) g-n
Yellow (Y0-Yn) y-n
Blue (B0-Bn) b-n
Cboot can currently be used to power and boot, power (on, off, cycle), or boot collections or individual devices, ranges of devices etc. A collection is defined as a grouping of devices as defined by the persistant object store (database) of the system.
Note: when powering or booting groups of racks, if the rack
contains a leader/mother/boohost/diag-node, that node will
not be operated on, unless the --leader switch is specified,
in which case only the leader of the specified nodes will be acted
upon. Also if that node is not functional the
operation will not be executed. The cboot command will try to
do the "right" thing when options are mixed. For example if
multiple racks and multiple nodes are both specified, action
will take place on those nodes specified on the full range
of racks specified (see examples). Also if a more global option
like --vm is used in addition to a rack specification, only
the nodes in both the vm and the specified rack will be operated
on (again see examples)
To power and then boot a group of equipment called a rack, specify the rack designation by name. For example a rack that is labeled "R0" (r-0) will be specified as follows:
cboot -r 0
The above command will first power and then boot rack r-0.
To power and boot rack T0 (t-0):
cboot -t 0
To power and boot a range of racks in the R series, specifically racks 0 through 3 inclusively.
cboot -r 0-3
To power and boot a range of racks that includes different letter designated racks, specifically racks in the R series 0-2, and racks in the T series 4-9.
cboot -r 0-2 -t 4-9
To specifiy multiple racks that are not inclusive, for example T series racks 4, 6, 7, and 9.
cboot -t 4,6,7,9
The same commands can be used to only power or boot the designated equipment. For example to power (cycle by default) T series racks 0 through 4, simply add the "--power" switch.
cboot -t 0-4 --power
To power "off" the same racks.
cboot -t 0-4 --power off
To power the same racks back on.
cboot -t 0-4 --power on
Now lets assume that you have power-cycled T series racks 0-4, and you want to avoid having to power-cycle them again before booting. Use the "--boot" switch to specify boot only.
cboot -t 0-4 --boot
The same actions are also supported on specific devices or nodes within rack groups. For example to power and then boot node 4 on T series rack 8.
cboot -t 8 -n 4
Ranges are also supported for the nodes only, for example to power and boot nodes 4 through 9 on T series rack 8.
cboot -t 8 -n 4-9
Non-inclusive ranges are also supported, for example to power and boot nodes 4, 6 and 8 on T series rack 8.
cboot -t 8 -n 4,6,8
Again the same commands can be used to only power (cycle by defualt) the nodes by adding the "--power" switch, or only boot the nodes with the "--boot" switch.
cboot -t 8 -r 7 -n 4,6,8
This command will first power cycle and then boot nodes 4, 6 and 8 in racks t-8 ane r-7.
cboot --vm default --power
This command will power cycle all of the nodes in the vm named "default". Work will be allocated to the leaders of the nodes.
cboot --vm default --power -t 9
This command will only power nodes that are in the vm named "default" that are in rack t-9.
cboot --power node.n-0.t-0
cboot also supports specifying a device name on the command line to act on.
Note: cboot will take the intersection of the devices specified in the
case where more than one way to designate devices is used. For instance
if the user specifies the --vm switch in conjunction with specifying
a device by using --t 0 --n 0 the only device that will be acted on
is node.n-0.t-0 provided it appears in the vm listed. If this device
is not a member of the listed vm no action will occur. Likewise if
a user specifies a node on the command line along with either the
--vm switch or the parameter method of device specification only
nodes that are in both specifications will be acted upon. If all three
methods are used, only devices that are in all groups specified are
acted upon.
Ex.
cboot -t 0 -n 0 --vm default
Only node node.n-0.t-0 will be acted upon, provided it is in the default vm.
cboot --vm default node.n-0.t-0
Same thing
cboot --vm default node.n-0.t-0 -t 0 -n 0
Again same thing
cboot --vm default node.n-1.t-0 -t 0 -n 0
No nodes will be acted upon because there is no intersection.
Use this to form creative ways to control the system.
/<install_dir>/bin/power
/<install_dir>/bin/boot
/<install_dir>/site/cboot
/<install_dir>/lib/Parse.pm
/<install_dir>/lib/Local.pm
The power and boot documents.
Questions, comments, bugs should be sent to jhlaros@sandia.gov