sesmgr(1m) sesmgr(1m)
sesmgr - Fibre Channel Drive Enclosure Status/Configuration Command Line
Interface
The command line syntax of sesmgr takes multiple flavors depending on the
option.
sesmgr getversion
sesmgr poll
sesmgr topology [-c channel-ID [-c channel-ID] . . .
sesmgr getstatus [-v] [-c channel-ID] [-e enclosure-ID] . .
sesmgr remove -t target-ID [-t target-ID] . . .
sesmgr ledon -t target-ID [-t target-ID] . . .
sesmgr ledoff -t target-ID [-t target-ID] . . .
sesmgr bypass -t target-ID -p A|B [-t target-ID -p A|B] . . .
sesmgr unbypass -t target-ID -p A|B [-t target-ID -p A|B] . . .
sesmgr insert -e enclosure-ID [-e enclosure-ID] . . .
sesmgr alarmon -e enclosure-ID [-e enclosure-ID] . . .
sesmgr alarmmute -e enclosure-ID [-e enclosure-ID] . . .
sesmgr alarmoff -e enclosure-ID [-e enclosure-ID] . . .
The sesmgr command is a user front end to sesdaemon which collects status
of all supported fibre channel enclosures and allows for configuration
changes.
getversion This command will report the version level of sesmgr.
poll This command will request that the daemon re-scan the SAN
for its configuration and also update the status of all
known devices. Typically this should only need to be done
if the SAN configuration is manually changed outside of
the sesmgr command set.
topology This command will print the daemon's understanding of the
topology of the SAN. It will print all of the drive
target's worldwide names, port id's, port numbers, and if
the controller is in loop mode, the name the device is
known by. A list of all known drive targets which
sesdaemon is not able to manage is also printed. This
list would include RAIDS (as its not easily possible to
distinguish them from JBODS), targets in unsupported
enclosures, and targets in enclosures which do not have
working devices in the SES capable slots for the ports
reachable by the host running sesdaemon. Targets which
are not disk drives are ignored by sesdaemon.
Page 1
sesmgr(1m) sesmgr(1m)
getstatus This command reports status of supported fibre channel
drive enclosures.
For this option, -v (verbose) alone will print a detailed
status of all enclosures, -v with specified channel
numbers will print a detailed status for the enclosures on
the listed channels. If enclosure arguments are specified
then a detailed status will be printed for only those
enclosures (and the -v flag is unnecessary).
If the -v flag is not used, and no arguments are
specified, then overall summaries of all fibre channel
controllers will be printed. Or only for the specified
channels if channel arguments are used.
remove This command will allow for physical removal of a drive.
I.E. it will spin-down the specified disk drive, and
prompt the user for removal, then after all the removes
have taken place it will SOP_SCAN the effected
controllers.
Note: This command issues the spindown command to the
drive immediately which locks the drive heads. However,
the drive requires additional time to stop rotating and
will still be turning when the exchange prompt is issued.
insert This command will allow for inserting a drive into a
specified enclosure. I.E. it will prompt for insertion,
and after all the inserts have taken place, it will spin
it up the drives, SOP_SCAN the effected controllers, and
call ioconfig to update the hardware graph.
ledon This command will turn "ON" the indicator "LED" associated
with the specified disk device.
ledoff This command will turn "OFF" the indicator "LED"
associated with the specified disk device.
bypass This command will perform an enclosure bypass to remove
the specified disk on the specified port from the fibre
channel loop.
unbypass This command will perform an enclosure unbypass to add the
specified disk on the specified port onto the fibre
channel loop.
alarmon This command will sound the alarm on the specified
enclosure.
alarmoff This command will turn off the alarm on the specified
enclosure provided that the enclosure is clear of any
other hardware conditions which also result in the alarm
Page 2
sesmgr(1m) sesmgr(1m)
sounding.
alarmmute This command will mute the enclosure's alarm. Generally
this will result in a periodic reminder tone instead of
totally silencing it.
COMMAND LINE OPTIONS
-c channel-ID Use this option to limit a status command to a particular
fibre channel controller.
-e enclosure-ID
Use this option to specify a particular enclosure.
Enclosures are identified by a 64 bit hex number similar
to a world wide name. (Don't forget the 0x prefix on the
arguments for this option.)
-t target-ID Use this option to direct an action at a particular drive
or set of drives. Targets are identified by their world
wide name, a 64 bit hex number unique to each drive.
(Don't forget the 0x prefix.) If the target is operating
on a private loop, it is also possible to specify it by
its sc* name in /hw/scsi (eg sc4d65l0).
Obtaining the revision level of the instance of sesmgr running locally
sesmgr getversion
sample output:
sesdaemon version 1.0, Thu May 11 18:46:09 PDT 2000
Obtaining a brief channel status summary:
sesmgr getstatus
sample output:
CHANNEL | DRIVES POWER LCC FANS TEMP ALARM
--------+--------------------------------------------------
2 | NONE NONE NONE NONE NONE NONE
3 | OK FAIL OK OK OK ON
4 | OK OK OK OK OK NONE
The example shows that channel 3 has one or more enclosures reporting a
power supply fault and an alarm is sounding. Channel 2 shows "NONE" which
indicates a few possibilities.
Page 3
sesmgr(1m) sesmgr(1m)
o There are no recognized fibre channel enclosures connect to the
channel.
o There is a communication problem on the channel.
o "NONE" can also mean that none of the recognized enclosures support
that feature. As on channel 4, the enclosures do not include an
alarm.
Obtaining a detailed status of an enclosure:
sesmgr getstatus -e 0x50050cc00000205c
sample output:
Enclosure ID: 0x50050cc00000205c
Vendor ID: XYRATEX
Product ID: SS-1202-FCAL
FRU's (Inst/Max): 11 of 12 drives
2 of 3 power supplies
1 of 2 recognized lcc's
Last successful status poll: Tue May 16 17:19:17 2000
+---------------------------+ Slot Worldwide Name (*=SES)
| 1/1 | 2/1 | 3/1 | 4/1 | 1/1 0x20000020374bab85
| OK | OK | OK | OK | 2/1 0x20000020374bb5e2
| | | | | 3/1 0x20000020374bb504
+---------------------------+ 4/1 0x20000020374b17ee
| 1/2 | 2/2 | 3/2 | 4/2 | 1/2 0x20000020374baf03
| OK | OUT | OK | OK | 2/2 0x0
| | | | | 3/2 0x20000020374bafdc
+---------------------------+ 4/2 0x20000020374b194f
| 1/3 | 2/3 | 3/3 | 4/3 | 1/3 0x20000020374bb280*
| OK | OK | OK | OK | 2/3 0x20000020374bb007
| | | | | 3/3 0x20000020374bafe1
+---------------------------+ 4/3 0x20000020374bb271*
FAN0 OK FAN1 OK FAN2 OUT
PS0 OK PS1 OK PS2 OUT
TEMP OK 29 C ALARM OK
The example shows that slot 2/2 does not have a drive in it, and both fan
2 and power supply 2 are not installed. Also this enclosure either has
only one fibre channel loop connection or else only one is visible
(connected) to the host.
Illuminating the LEDs associated with disk 1/1 and 2/1:
sesmgr ledon -t 0x20000020374bab85 -t 0x20000020374bb5e2
Page 4
sesmgr(1m) sesmgr(1m)
LIMITATIONS
Commands which change the enclosure status will not be reflected by
getstatus until the next status poll executes (or until a poll command is
run).
Currently the only supported enclosures are the Xyratex 12 slot and
Clariion 10 slot enclosures.
Not all enclosures will support all commands or types of FRU's. For
example, the Xyratex 12 slot enclosure does not support enclosure bypass
and unbypass. And the Clariion enclosure does not have an alarm.
The ledon and ledoff commands may indicate differently depending on the
enclosure's implementation. For example, the Xyratex 12 will slowly
blink the activity led when its "on", and leave it lit solidly when
"off". Clariion enclosures will solidly indicate on the fault LED.
Removing a drive can send a controller into error recovery mode as it
tries to find the target and bring it back online.
It may take a while for SOP_SCAN and ioconfig to complete depending on
how large and how busy the SAN is.
When a device is specified by its sc* loop name, the code will find that
device. However commands to that device may use alternate paths.
On poll updates its possible that failed equipment may disappear and no
longer be shown as failed. Failed targets can be dropped from the target
list, but they should still be printed in their enclosure slot by
getstatus. Enclosures which are no longer reachable (and their targets)
may be dropped entirely. This is due to a configuration update
assumption that targets which do not respond were intentionally moved.
bypass and unbypass can fail if the commands can not be issued on the
same loop that the specified device's port is on.
If a drive is bypassed so that the system no longer sees an active
target, and the daemon is restarted, the target may no longer be seen.
In this situation it will not be possible to issue an unbypass command to
the target.
In the event that an enclosure hasnt been heard from during a period
three times sesdaemn's poll period, getstatus will show OLD for the
status fields. If an enclosure has been disconnected, use the poll
command to update the configuration and status information. More
typically, OLD indicates that there is a communications problem.
Depending on the nature of the problem, error recovery can hold off the
sequential polling of other controllers. Be advised however that OLD
will NOT be listed if the enclosures for a controller can be reached via
alternate paths.
In order to run sesmgr under TRIX, the user must have a MAC label of
Page 5
sesmgr(1m) sesmgr(1m)
dblow and a capability of CAP_PRIV_PORT.
DIFFERENCES FROM FCCLI [Toc] [Back] Although the sesdaemon/sesmgr utilities are designed to be similar in
look and functionality, there are some important differences. Here are
some of these:
Enclosure ID's and target ID's are now 64 bit hex integers.
insert now takes an enclosure ID, not a target ID.
An argument id is now required before each one (eg -t) to tell what type
it is.
Target lists are no longer restricted to an individual controller.
bypass and unbypass are enclosure commands only. Loop commands are not
attempted.
/usr/sbin/sesmgr
ioconfig(1M), scsiha(7), sesdaemon(1M)
PPPPaaaaggggeeee 6666 [ Back ]
|
