Ignite question

[Fabian Briseño]

Hello.
I have the following situation.

I have a L-1000 server with 2 18GB internal disks, one of these disks is damaged (NO_HW),
when I try to do an ignite the following message is displayed.

/var/opt/ignite/recovery/2006-04-06,22:44/system_cfg vg00
lssf: Must specify a special_file
usage: lssf special_file ...
save_config: error - unknown disk type for , not SCSI or HPFL
save_config: error - cannot determine root disk
ERROR: /opt/ignite/bin/save_config failed

I suppose this is because of the failed disk.
I have used the following instruction for doing the ignite all of them give the same result (see above).

make_tape _recovery -Av

make_tape_recovery -x inc_entire=/dev/dsk/vg00 -A


make_tape_recovery -x inc_entire=/dev/dsk/c1t2d0 (this is the hardware path of the disk that is still good)

In these two internal disk the OS is installed is there anyway to force ignite to do a backup on the disk that is still good ?.

thanks in advance

[Fabio Ettore · Accepted solution]

Hi Fabian,

is the damaged disk just a mirror? If so then you can try to reduce the mirror from there (probably you should use the -k option on lvreduce commands). Once reduced try again the make_tape_recovery.

Othrwise I see it difficult to backup only the good disk. I also hope you already have a good backup.

Good luck....

Best regards,
Fabio

[Fabio Ettore]

Hi Fabian,

is the damaged disk just a mirror? If so then you can try to reduce the mirror from there (probably you should use the -k option on lvreduce commands). Once reduced try again the make_tape_recovery.

Otherwise I see it difficult to backup only the good disk, I don't think it's possible, never heard that before. I also hope you already have a good backup.

The correct make_tape_recovery should be

make_tape_recovery -v -x inc_entire=vg00

-A option is currently available on make_tape_recovery but it's not to obtain the same result as -x inc_entire=vg00.


Good luck....

Best regards,
Fabio

[Mridul Shrivastava]

if u execute lvdisplay for vg then u'll come to know that on which disk there are stale PEs, then u can try reducing the each LV. or after that execute lvdisplay command with -k option so we'll know the pvkey no. for the disk (make a note for disk which has stale PEs).
Than excute lvreduce command using that pvkey value along with the -k option.
Then u go ahead with taking backup, I'll recommend u too replace the failed hard disk as soon as possible to avoid any disaster.

[Prashanth.D.S]

Hi Briseno,

Pls check if you get the following error message after running #make_tape _recovery -Av

lvdisplay: Warning: couldn't query all of the physical volumes.
lvdisplay: Warning: couldn't query physical volume "/dev/dsk/c3t0d5":
The specified path does not correspond to physical volume attached to
this volume group

If yes

Solution:

The errors here are not indicative of a problem with ignite and make_recovery.
They are a result of an underlying problem with the Logical Volume Manager
(LVM). Make_recovery uses the /opt/ignite/bin/save_config script to read
the LVM configuration information and create a configuration file for the
make_recovery tape archive. A look at the /etc/lvmtab file shows that some
physical volumes in the volume group vg00 that had been removed and changed
were still in the file.

# mv /etc/lvmtab /etc/lvmtab.old
# vgscan -v

This rebuilt the lvmtab file with the correct physical volumes present and
resolved the LVM command type errors (lvdisplay, vgdisplay, etc). We then
copied and ran the save_config script manually to check the results:

# cp /opt/ignite/bin/save_config /tmp

To assist us in debugging on we vi the script and on the second line of the
script we insert: set -x

Now run it and dump all of the standard and error output to a file:

# /tmp/save_config >/tmp/sc_info 2>&1
# more sc_info
...text truncated
+ find_root_disk
+ [[ = /dev/root ]]
+ + disk_to_path
lssf: Must specify a special_file
usage: lssf special_file ...
save_config: error - unknown disk type for , not SCSI or HPFL
root_hardware_path=
+ print -u4 init _hp_root_disk=""
+ [[ -z ]]
+ fatal cannot determine root disk
save_config: error - cannot determine root disk

The script still cannot determine the root disk drive. Using lvlnboot also
fails:

# lvlnboot -v
Boot Definitions for Volume Group /dev/vg00:
Physical Volumes belonging in Root Volume Group:
/dev/dsk/c1t1d3 (3/0/0.1.3) -- Boot Disk
Unable to query the I/O interface: Parameter error.
lvlnboot: Unable to print reliable information.

By recovering the missing links to all of the logical volumes specified in the
Boot Data Reserved Area and update the Boot Data Reserved Area of each bootable
physical volume in the volume group we can get the lvlnboot command to work:

# lvlnboot -R /dev/vg00
Volume Group configuration for /dev/vg00 has been saved in
/etc/lvmconf/vg00.conf

# lvlnboot -v
Boot Definitions for Volume Group /dev/vg00:
Physical Volumes belonging in Root Volume Group:
/dev/dsk/c1t1d3 (3/0/0.1.3) -- Boot Disk
Boot: lvol1 on: /dev/dsk/c1t1d3
Root: lvol3 on: /dev/dsk/c1t1d3
Swap: lvol2 on: /dev/dsk/c1t1d3
Dump: lvol2 on: /dev/dsk/c1t1d3, 0


This will rebuild the /etc/lvmconf/vg00.conf file with the correct information for
the root drive and make_recovery.

[Fabian Briseño]

thanks for the replys guys.

I'll try your advice, as soon as possible.

I must say that this server is not critical and in fact we will be moving the application there in a week or so to a new server.



Thank you again.