Performing maintenance when booted from cdrom

东方蜘蛛

Introduction
If the system has to be booted from a cdrom or from the network ("boot cdrom" or "boot net") in order to perform maintenance, the operator needs to adjust for the existence of a mirrored operating system. Because these alternate boot devices do not include the drivers necessary for veritas volume manager, they cannot be used to operate on veritas volumes. This raises subtle issues addressed below.
Typically, the administrator is often under pressure while performing these types of maintenance. Because simple mistakes at this stage can render the system unusable, it is important that the process be well documented and tested prior to using it in production.
In the example below, the server pegasus has two internal disks (c0t0d0 and c0t1d0) under VxVM control. The operating system is mirrored between the two devices. Assume that the administrator has forgotten the root password on this server, and needs to boot from cdrom in order to edit the shadow file.
Boot from cdrom and update shadow file

Password forgotten! Insert the Solaris operating system CD into the cdrom drive and boot from it into single-user mode:
pegasus console login: root
Password:
Login incorrect
Oct 25 17:33:44 pegasus login: REPEATED LOGIN FAILURES ON /dev/console, root
pegasus console login: root
Password:
Login incorrect
pegasus console login:
Type  'go' to resume
ok boot cdrom -s
Resetting ...
screen not found.
Can't open input device.
Keyboard not present.  Using ttya for input and output.
Sun Ultra 30 UPA/PCI (UltraSPARC-II 296MHz), No Keyboard
OpenBoot 3.27, 512 MB memory installed, Serial #9377973.
Ethernet address 8:0:20:8f:18:b5, Host ID: 808f18b5.
Initializing Memory Rebooting with command: boot cdrom -s
Boot device: /pci@1f,4000/scsi@3/disk@6,0:f  File and args: -s
SunOS Release 5.8 Version Generic_108528-07 64-bit
Copyright 1983-2001 Sun Microsystems, Inc.  All rights reserved.
Configuring /dev and /devices
Using RPC Bootparams for network configuration information.
INIT: SINGLE USER MODE
#

Fsck and mount the root disk's "/" partition in order to edit the /etc/shadow file:
# fsck -y /dev/rdsk/c0t0d0s0
# mount /dev/dsk/c0t0d0s0 /a

Remove the encrypted password from the /a/etc/shadow file:
# TERM=vt100; export TERM
# vi /a/etc/shadow
For example, if the entry for the root user looks like the following:
root:NqfAn3tWOy2Ro:6445::::::
Change it so that is looks as follows:
root::6445::::::

Unmount the root filesystem:
# cd /; umount /a
"Zero out" the private and public regions of the rootmirror disk

Since we've updated just one half of the mirror device, one needs to convince Veritas volume manager to not make use of the other half of the mirror. To do this, we "zero out" the public and private regions of rootmirror c0t1d0. When Veritas boots and cannot detect the private region on c0t1d0, it will assume that the disk has failed:
# format
Searching for disks...done
AVAILABLE DISK SELECTIONS:
       0. c0t0d0
          /pci@1f,4000/scsi@3/sd@0,0
       1. c0t1d0
          /pci@1f,4000/scsi@3/sd@1,0
Specify disk (enter its number): 1
selecting c0t1d0
[disk formatted]
. . .
partition> p
Current partition table (original):
Total disk cylinders available: 5266 + 2 (reserved cylinders)
Part      Tag    Flag     Cylinders        Size            Blocks
  0       root    wm       1 - 3995        6.40GB    (3995/0/0) 13423200
  1       swap    wu    3996 - 4620        1.00GB    (625/0/0)   2100000
  2     backup    wm       0 - 5265        8.44GB    (5266/0/0) 17693760
  3          -    wu       0 -    0        1.64MB    (1/0/0)        3360
  4          -    wu       1 - 5265        8.44GB    (5265/0/0) 17690400
  5 unassigned    wm       0               0         (0/0/0)           0
  6        var    wm    4621 - 5245        1.00GB    (625/0/0)   2100000
  7 unassigned    wm       0               0         (0/0/0)           0
partition> 3
Part      Tag    Flag     Cylinders        Size            Blocks
  3          -    wu       0 -    0        1.64MB    (1/0/0)        3360
Enter partition id tag[unassigned]:
Enter partition permission flags[wu]:
Enter new starting cyl[0]:
Enter partition size[3360b, 1c, 1.64mb, 0.00gb]: 0
partition> 4
Part      Tag    Flag     Cylinders        Size            Blocks
  4          -    wu       1 - 5265        8.44GB    (5265/0/0) 17690400
Enter partition id tag[unassigned]:
Enter partition permission flags[wu]:
Enter new starting cyl[1]: 0
Enter partition size[17690400b, 5265c, 8637.89mb, 8.44gb]: 0
partition> p
Current partition table (unnamed):
Total disk cylinders available: 5266 + 2 (reserved cylinders)
Part      Tag    Flag     Cylinders        Size            Blocks
  0       root    wm       1 - 3995        6.40GB    (3995/0/0) 13423200
  1       swap    wu    3996 - 4620        1.00GB    (625/0/0)   2100000
  2     backup    wm       0 - 5265        8.44GB    (5266/0/0) 17693760
  3 unassigned    wu       0               0         (0/0/0)           0
  4 unassigned    wu       0               0         (0/0/0)           0
  5 unassigned    wm       0               0         (0/0/0)           0
  6        var    wm    4621 - 5245        1.00GB    (625/0/0)   2100000
  7 unassigned    wm       0               0         (0/0/0)           0
partition> label
Ready to label disk, continue? y
partition> quit
Boot from the primary boot disk

Return to the ok prompt and boot from the primary boot device:
# stop-a
ok boot disk
Resetting ...
screen not found.
Can't open input device.
Keyboard not present.  Using ttya for input and output.
Sun Ultra 30 UPA/PCI (UltraSPARC-II 296MHz), No Keyboard
OpenBoot 3.27, 512 MB memory installed, Serial #9377973.
Ethernet address 8:0:20:8f:18:b5, Host ID: 808f18b5.
Initializing MemoryRebooting with command: boot disk
Boot device: /pci@1f,4000/scsi@3/disk@0,0  File and args:
SunOS Release 5.8 Version Generic_108528-16 64-bit
Copyright 1983-2001 Sun Microsystems, Inc.  All rights reserved.
Starting VxVM restore daemon...
VxVM starting in boot mode...
/usr/sbin/prtconf: getexecname() failed
vxvm:vxconfigd: WARNING: Detaching plex rootvol-02 from volume rootvol
vxvm:vxconfigd: WARNING: Disk rootmirror in group rootdg: Disk device not found
configuring IPv4 interfaces: hme0.
Hostname: pegasus
VxVM starting special volumes ( swapvol rootvol var )...
VxVM general startup...
dumpadm: no swap devices could be configured as the dump device
The system is coming up.  Please wait.
starting rpc services: rpcbind done.
Setting netmask of hme0 to 255.255.255.0
Setting default IPv4 interface for multicast: add net 224.0/4: gateway pegasus
Starting sshd...
This platform does not support both privilege separation and compression
Compression disabled
syslog service starting.
savecore: no dump device configured
savecore: no dump device configured
dumpadm: no swap devices could be configured as the dump device
Oct 25 17:44:21 pegasus savecore: no dump device configured
Print services started.
/dev/bd.off: not a serial device.
volume management starting.
  No VVR license installed on the system; vradmind not started.
  No VVR license installed on the system; in.vxrsyncd not started.
The system is ready.

Login to the system (without a password) and confirm that VxVM has determined that the rootmirror disk has failed.
pegasus console login: root
Last login: Fri Oct 25 17:02:05 from rambler.wakefie
Oct 25 17:44:46 pegasus login: ROOT LOGIN /dev/console
Sun Microsystems Inc.   SunOS 5.8       Generic February 2000
You have new mail.
# vxdisk list
DEVICE       TYPE      DISK         GROUP        STATUS
c0t0d0s2     sliced    rootdisk     rootdg       online
c0t1d0s2     sliced    -            -            error
-            -         rootmirror   rootdg       failed was:c0t1d0s2
# vxprint -ht
Disk group: rootdg
DG NAME         NCONFIG      NLOG     MINORS   GROUP-ID
DM NAME         DEVICE       TYPE     PRIVLEN  PUBLEN   STATE
RV NAME         RLINK_CNT    KSTATE   STATE    PRIMARY  DATAVOLS  SRL
RL NAME         RVG          KSTATE   STATE    REM_HOST REM_DG    REM_RLNK
V  NAME         RVG          KSTATE   STATE    LENGTH   READPOL   PREFPLEX UTYPE
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
SV NAME         PLEX         VOLNAME  NVOLLAYR LENGTH   [COL/]OFF AM/NM    MODE
DC NAME         PARENTVOL    LOGVOL
SP NAME         SNAPVOL      DCO
dg rootdg       default      default  0        1035555399.1025.pegasus
dm rootdisk     c0t0d0s2     sliced   3359     17690400 -
dm rootmirror   -            -        -        -        NODEVICE
v  rootvol      -            ENABLED  ACTIVE   13423200 ROUND     -        root
pl rootvol-01   rootvol      ENABLED  ACTIVE   13423200 CONCAT    -        RW
sd rootdisk-B0  rootvol-01   rootdisk 17690399 1        0         c0t0d0   ENA
sd rootdisk-02  rootvol-01   rootdisk 0        13423199 1         c0t0d0   ENA
pl rootvol-02   rootvol      DISABLED NODEVICE 13423200 CONCAT    -        RW
sd rootmirror-01 rootvol-02  rootmirror 0      13423200 0         -        NDEV
v  swapvol      -            ENABLED  ACTIVE   2100000  ROUND     -        swap
pl swapvol-01   swapvol      ENABLED  ACTIVE   2100000  CONCAT    -        RW
sd rootdisk-01  swapvol-01   rootdisk 13423199 2100000  0         c0t0d0   ENA
pl swapvol-02   swapvol      DISABLED NODEVICE 2100000  CONCAT    -        WO
sd rootmirror-02 swapvol-02  rootmirror 13423200 2100000 0        -        NDEV
v  var          -            ENABLED  ACTIVE   2100000  ROUND     -        fsgen
pl var-01       var          ENABLED  ACTIVE   2100000  CONCAT    -        RW
sd rootdisk-03  var-01       rootdisk 15523199 2100000  0         c0t0d0   ENA
pl var-02       var          DISABLED NODEVICE 2100000  CONCAT    -        WO
sd rootmirror-03 var-02      rootmirror 15523200 2100000 0        -        NDEV

Change the root password for the system:
# passwd root
New password: *******
Re-enter new password:
passwd (SYSTEM): passwd successfully changed for root
Replace the rootmirror with itself

Go through the standard process of replacing the c0t1d0 disk, without actually physically changing the disk device:
# vxdiskadm
Volume Manager Support Operations
Menu: VolumeManager/Disk
1      Add or initialize one or more disks
2      Encapsulate one or more disks
3      Remove a disk
4      Remove a disk for replacement
5      Replace a failed or removed disk
6      Mirror volumes on a disk
7      Move volumes from a disk
8      Enable access to (import) a disk group
9      Remove access to (deport) a disk group
10     Enable (online) a disk device
11     Disable (offline) a disk device
12     Mark a disk as a spare for a disk group
13     Turn off the spare flag on a disk
14     Unrelocate subdisks back to a disk
15     Exclude a disk from hot-relocation use
16     Make a disk available for hot-relocation use
17     Prevent multipathing/Suppress devices from VxVM's view
18     Allow multipathing/Unsuppress devices from VxVM's view
19     List currently suppressed/non-multipathed devices
20     Change the disk naming scheme
21     Get the newly connected/zoned disks in VxVM view
list   List disk information
?      Display help about menu
??     Display help about the menuing system
q      Exit from menus
Select an operation to perform: 4
Remove a disk for replacement
Menu: VolumeManager/Disk/RemoveForReplace
  Use this menu operation to remove a physical disk from a disk
  group, while retaining the disk name.  This changes the state
  for the disk name to a "removed" disk.  If there are any
  initialized disks that are not part of a disk group, you will be
  given the option of using one of these disks as a replacement.
Enter disk name [,list,q,?] list
Disk group: rootdg
DM NAME         DEVICE       TYPE     PRIVLEN  PUBLEN   STATE
dm rootdisk     c0t0d0s2     sliced   3359     17690400 -
dm rootmirror   -            -        -        -        NODEVICE
Enter disk name [,list,q,?] rootmirror
  The following volumes will lose mirrors as a result of this
  operation:
        rootvol swapvol var
  No data on these volumes will be lost.
  The requested operation is to remove disk rootmirror from disk group
  rootdg.  The disk name will be kept, along with any volumes using
  the disk, allowing replacement of the disk.
  Select "Replace a failed or removed disk" from the main menu
  when you wish to replace the disk.
Continue with operation? [y,n,q,?] (default: y)
  Removal of disk rootmirror completed successfully.
Remove another disk? [y,n,q,?] (default: n)
Volume Manager Support Operations
Menu: VolumeManager/Disk
1      Add or initialize one or more disks
2      Encapsulate one or more disks
3      Remove a disk
4      Remove a disk for replacement
5      Replace a failed or removed disk
6      Mirror volumes on a disk
7      Move volumes from a disk
8      Enable access to (import) a disk group
9      Remove access to (deport) a disk group
10     Enable (online) a disk device
11     Disable (offline) a disk device
12     Mark a disk as a spare for a disk group
13     Turn off the spare flag on a disk
14     Unrelocate subdisks back to a disk
15     Exclude a disk from hot-relocation use
16     Make a disk available for hot-relocation use
17     Prevent multipathing/Suppress devices from VxVM's view
18     Allow multipathing/Unsuppress devices from VxVM's view
19     List currently suppressed/non-multipathed devices
20     Change the disk naming scheme
21     Get the newly connected/zoned disks in VxVM view
list   List disk information
?      Display help about menu
??     Display help about the menuing system
q      Exit from menus
Select an operation to perform: 5
Replace a failed or removed disk
Menu: VolumeManager/Disk/ReplaceDisk
  Use this menu operation to specify a replacement disk for a disk
  that you removed with the "Remove a disk for replacement" menu
  operation, or that failed during use.  You will be prompted for
  a disk name to replace and a disk device to use as a replacement.
  You can choose an uninitialized disk, in which case the disk will
  be initialized, or you can choose a disk that you have already
  initialized using the Add or initialize a disk menu operation.
Select a removed or failed disk [,list,q,?] list
Disk group: rootdg
DM NAME         DEVICE       TYPE     PRIVLEN  PUBLEN   STATE
dm rootmirror   -            -        -        -        REMOVED
Select a removed or failed disk [,list,q,?] rootmirror
Select disk device to initialize [,list,q,?] list
DEVICE       DISK         GROUP        STATUS
c0t0d0       rootdisk     rootdg       online
c0t1d0       -            -            error
Select disk device to initialize [,list,q,?] c0t1d0
  The following disk device has a valid VTOC, but does not appear to have
  been initialized for the Volume Manager.  If there is data on the disk
  that should NOT be destroyed you should encapsulate the existing disk
  partitions as volumes instead of adding the disk as a new disk.
  Output format: [Device_Name]
  c0t1d0
Encapsulate this device? [y,n,q,?] (default: y) n
  c0t1d0
Instead of encapsulating, initialize? [y,n,q,?] (default: n) y
  The requested operation is to initialize disk device c0t1d0 and
  to then use that device to replace the removed or failed disk
  rootmirror in disk group rootdg.
Continue with operation? [y,n,q,?] (default: y)
  Replacement of disk rootmirror in group rootdg with disk device
  c0t1d0 completed successfully.
Replace another disk? [y,n,q,?] (default: n) n
# vxdisk list
DEVICE       TYPE      DISK         GROUP        STATUS
c0t0d0s2     sliced    rootdisk     rootdg       online
c0t1d0s2     sliced    rootmirror   rootdg       online
# vxprint -ht
Disk group: rootdg
DG NAME         NCONFIG      NLOG     MINORS   GROUP-ID
DM NAME         DEVICE       TYPE     PRIVLEN  PUBLEN   STATE
RV NAME         RLINK_CNT    KSTATE   STATE    PRIMARY  DATAVOLS  SRL
RL NAME         RVG          KSTATE   STATE    REM_HOST REM_DG    REM_RLNK
V  NAME         RVG          KSTATE   STATE    LENGTH   READPOL   PREFPLEX UTYPE
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT    NCOL/WID MODE
SD NAME         PLEX         DISK     DISKOFFS LENGTH   [COL/]OFF DEVICE   MODE
SV NAME         PLEX         VOLNAME  NVOLLAYR LENGTH   [COL/]OFF AM/NM    MODE
DC NAME         PARENTVOL    LOGVOL
SP NAME         SNAPVOL      DCO
dg rootdg       default      default  0        1035555399.1025.pegasus
dm rootdisk     c0t0d0s2     sliced   3359     17690400 -
dm rootmirror   c0t1d0s2     sliced   3359     17690400 -
v  rootvol      -            ENABLED  ACTIVE   13423200 ROUND     -        root
pl rootvol-01   rootvol      ENABLED  ACTIVE   13423200 CONCAT    -        RW
sd rootdisk-B0  rootvol-01   rootdisk 17690399 1        0         c0t0d0   ENA
sd rootdisk-02  rootvol-01   rootdisk 0        13423199 1         c0t0d0   ENA
pl rootvol-02   rootvol      ENABLED  STALE    13423200 CONCAT    -        WO
sd rootmirror-01 rootvol-02  rootmirror 0      13423200 0         c0t1d0   ENA
v  swapvol      -            ENABLED  ACTIVE   2100000  ROUND     -        swap
pl swapvol-01   swapvol      ENABLED  ACTIVE   2100000  CONCAT    -        RW
sd rootdisk-01  swapvol-01   rootdisk 13423199 2100000  0         c0t0d0   ENA
pl swapvol-02   swapvol      DISABLED RECOVER  2100000  CONCAT    -        WO
sd rootmirror-02 swapvol-02  rootmirror 13423200 2100000 0        c0t1d0   ENA
v  var          -            ENABLED  ACTIVE   2100000  ROUND     -        fsgen
pl var-01       var          ENABLED  ACTIVE   2100000  CONCAT    -        RW
sd rootdisk-03  var-01       rootdisk 15523199 2100000  0         c0t0d0   ENA
pl var-02       var          DISABLED RECOVER  2100000  CONCAT    -        WO
sd rootmirror-03 var-02      rootmirror 15523200 2100000 0        c0t1d0   ENA
#
Once the replacement process completes (monitor it via the vxtask list command), operating system redundancy has been restored


本文来自ChinaUnix博客，如果查看原文请点：http://blog.chinaunix.net/u/524/showart_87421.html