scsifo(1m)

NAME
     scsifo - Show Failover status and initiate Failover.


SYNOPSIS
     The command line syntax of scsifo takes two flavors depending on the use.

     scsifo -d
     scsifo -s device-name
     scsifo -t device-name

DESCRIPTION
     The first form is used to dump failover status, while the second and
     third are used to initiate failover. The device-name required when
     performing a failover should be that of the a disk (raw or block) or a
     devscsi device associated with the primary path of the group the failover
     is being attempted on.  The difference between the -s and -t options is
     that the former will choose paths that do not require a controller
     tresspass in preference to paths that do, while the latter will choose
     the next path in the list and automatically issue a tresspass command to
     it.  For disk subsystems that do not have a lun ownership property, the
     options are identical.


EXAMPLES
     Dumping the current failover status:

     The following example shows a single SCSI RAID configured with 8 LUNs.

       scsifo -d

       sample output:

       Group 0:
         [P] sc4d2l7 (587)
         [ ] sc5d2l7 (790)
       Group 1:
         [P] sc4d2l6 (579)
         [ ] sc5d2l6 (782)
       Group 2:
         [P] sc4d2l5 (571)
         [ ] sc5d2l5 (774)
       Group 3:
         [P] sc4d2l4 (563)
         [ ] sc5d2l4 (766)
       Group 4:
         [P] sc4d2l3 (555)
         [ ] sc5d2l3 (758)
       Group 5:
         [P] sc4d2l2 (547)
         [ ] sc5d2l2 (750)
       Group 6:
         [P] sc4d2l1 (539)
         [ ] sc5d2l1 (742)
       Group 7:
         [ ] sc4d2l0 (531)
         [P] sc5d2l0 (734)

     The current primary path is indicated by a [P] in the left column. Each
     LUN is shown as an independent failover group with two paths. For all but
     LUN 0, the primary path is accessed via SCSI bus 4. For LUN 0, the
     primary path is accessible via SCSI bus 5.

     Initiating failover:

     The following example shows the initiation of failover of Group 5 (see
     above example). The primary path will migrate from SCSI bus 4 to SCSI bus
     5. Note that the devscsi device name will be used for initiating the
     failover.  It is possible to initiate failover using the full canonical
     name, e.g.,
     /hw/module/1/slot/io3/mscsi/pci/2/scsi_ctlr/0/target/2/lun/2/scsi.

       scsifo -s sc4d2l2

       sample output:

       New primary path /hw/module/1/slot/io3/mscsi/pci/3/scsi_ctlr/0/target/2/lun/2/scsi (869) (sc5d2l2)

     Another dump of the failover status shows that LUN 2 (group 5) has in
     fact migrated.

       scsifo -d

       sample output:

       Group 0:
         [P] sc4d2l7 (587)
         [ ] sc5d2l7 (790)
       Group 1:
         [P] sc4d2l6 (579)
         [ ] sc5d2l6 (782)
       Group 2:
         [P] sc4d2l5 (571)
         [ ] sc5d2l5 (774)
       Group 3:
         [P] sc4d2l4 (563)
         [ ] sc5d2l4 (766)
       Group 4:
         [P] sc4d2l3 (555)
         [ ] sc5d2l3 (758)
       Group 5:
         [ ] sc4d2l2 (547)
         [P] sc5d2l2 (750)
       Group 6:
         [P] sc4d2l1 (539)
         [ ] sc5d2l1 (742)
       Group 7:
         [ ] sc4d2l0 (531)
         [P] sc5d2l0 (734)


FILES
     /usr/sbin/scsifo
     /dev/rdsk/*
     /dev/dsk/*
     /dev/scsi/*
     /etc/init.d/failover
     /etc/failover.conf
     /var/sysgen/master.d/failover

SEE ALSO
     failover(7M).

NOTES
     On bootup, if only a single path is found for a failover group and it's
     not the primary, auto failover may be performed depending on the setting
     of the kernel tunable variable fo_auto_failover_on_start. See
     /var/sysgen/master.d/failover.  This does not apply to devices defined
     within the /etc/failover.conf configuration file.