Swap logical volume size beyond the IODC max address.

[RAC_1]

I am preparing the third (static) copy of the boot disk. I am preparing vgroot VG for it. All goes OK, but I run intoproblem with command.
lvlnboot -s /dev/vgroot/lvol2. I get error as follows.

vlnboot: Unable to configure swap logical volume.
Swap logical volume size beyond the IODC max address.

The one thing that I do not understand is that primary root disk (under vg00) has same setup and has following lvlnboot -v vg00 output.

Boot Definitions for Volume Group /dev/vg00:
Physical Volumes belonging in Root Volume Group:
/dev/dsk/c4t6d0 (10/0.6.0) -- Boot Disk
/dev/dsk/c4t10d0 (10/0.10.0)
Boot: lvol1 on: /dev/dsk/c4t6d0
Root: lvol3 on: /dev/dsk/c4t6d0
Swap: lvol2 on: /dev/dsk/c4t6d0
Dump: lvol2 on: /dev/dsk/c4t6d0, 0

But, when setting /dev/vgroot/lvol2 as swap I get above mentioned error. I am on 11.11 running on K580. I am not very up to date on patches. But the fact that vg00 - lvol2 is already set to swap device without above error message, then why error is coming for /dev/vgroot/lvol2. All lvols created in exactly the same way as vg00 lvols. Exact same sizes, exact same policies. Outpus as follows.

--- Logical volumes ---
LV Name /dev/vg00/lvol2
VG Name /dev/vg00
LV Permission read/write
LV Status available/syncd
Mirror copies 0
Consistency Recovery MWC
Schedule parallel
LV Size (Mbytes) 4096
Current LE 1024
Allocated PE 1024
Stripes 0
Stripe Size (Kbytes) 0
Bad block off
Allocation strict/contiguous
IO Timeout (Seconds) default



--- Logical volumes ---
LV Name /dev/vgroot/lvol2
VG Name /dev/vgroot
LV Permission read/write
LV Status available/syncd
Mirror copies 0
Consistency Recovery MWC
Schedule parallel
LV Size (Mbytes) 4096
Current LE 1024
Allocated PE 1024
Stripes 0
Stripe Size (Kbytes) 0
Bad block off
Allocation strict/contiguous
IO Timeout (Seconds) default

The disk types used in vg00 and vgroot is same.
vendor: SEAGATE
product id: ST19171W
type: direct access
size: 8891556 Kbytes
bytes per sector: 512

I read some documents in knowledge base that said, it;s on account of limitation of 2GB for swap. But the fact that vg00-lvol2-swap is already set without this problem.

Thank you in Advance.

[Warren_9]

hi,

base on http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=800332&admit=-682735245+1146036077734+28353475

maybe you need to upragde the firmware.

GOOD LUCK!!

[RAC_1]

Warren,

As I informed. I already have lvol2 set as swap on vg00. So why there should be a problem with lvlnboot -s /dev/vgroot/lvol2

[Senthil Kumar .A_1]

Hi RAC,

I think the issue is due to the fact that, you are trying to set your primary swap as the logical volume that does not belong to your current "Boot disk". The only way you could have 2 or 3 LV's displayed under "Swap: " section of lvlnboot -v is when you mirror the swap. Here you are getting this error because you are trying to set a logical volume which is housed in a different disk other than boot disk. I doubt whether this activity is possible. Please wait for experienced admin's before taking my opinion. May be they can comment on my thought.

I haven't tried this though, I believe this activity could be possible if you boot from statically copied disk through "LVM Maintenece mode", because.. remember the system does not still know whats the primary swap to use when booted from this static copy of the disk. If you are able to boot successfully , then you can run, "lvlnboot" command and set the "root", "boot" , "swap" and "dump" sections in the boot area for the statically copied disk.

Regards,
Senthil Kumar .A

[Senthil Kumar .A_1]

Hi RAC,

Check the response of "stefan", his reply seems to be on the same line...

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=800332&admit=-682735245+1146046470400+28353475

Regards,
Senthil Kumar .A

[RAC_1]

I have done this on different system. All I am doing is preparing the static copy of boot disk. (boot disk is mirrored, but it's matter here)

It works on different system and I have done this quite a few times. It is just that this server gives error. Also, if IODC error has something to with IODC firmware, then we should have encountered this error on pripary boot disk (under vg00) also, but we did not.

[Kent Ostby]

RAC -- Are the disks all on the same I/O bus?

The "firmware" in question that needs to be upgraded is for the I/O card not for the CPU so even if you're doing it elsewhere on the system, if you're not doing it on this card, there could still be an issue with firmware.

[James R. Ferguson]

Hi RAC:

Did you mirror your first three logical volumes exactly in order (1,2,3)? If not,, break the mirror and do these (at least) in that order and then issue the 'lvlnboot' command. As I recall, not ordering thusly will lead to the error you see.

Regards!

...JRF...

[RAC_1]

Kent,

All disks are not on same card. check the ioscan output. How do I check the firware version on particular I/O card? (with stm?)
Also is it possible to update the firmware on particular I/O card?

ioscan output. (Soemthing seems wrong here though.)

Class I H/W Path Driver S/W State H/W Type Description
=============================================================
disk 0 10/0.6.0 sdisk CLAIMED DEVICE SEAGATE ST19171W
/dev/dsk/c4t6d0 /dev/rdsk/c4t6d0
Class I H/W Path Driver S/W State H/W Type Description
==============================================================
disk 1 10/0.10.0 sdisk CLAIMED DEVICE SEAGATE ST19171W
/dev/dsk/c4t10d0 /dev/rdsk/c4t10d0
Class I H/W Path Driver S/W State H/W Type Description
================================================================
disk 2 10/4/4.5.0 disc3 CLAIMED DEVICE SEAGATE ST19171W
/dev/dsk/c5t5d0 /dev/floppy/c5t5d0 /dev/rdsk/c5t5d0 /dev/rfloppy/c5t5d0
Class I H/W Path Driver S/W State H/W Type Description
=================================================================
disk 3 10/4/4.11.0 disc3 CLAIMED DEVICE SEAGATE ST19171W
/dev/dsk/c5t11d0 /dev/floppy/c5t11d0 /dev/rdsk/c5t11d0 /dev/rfloppy/c5t11d0

[Senthil Kumar .A_1]

Hi RAC,

Sorry for asking the question instead of indicating you a solution at this point.

I don't quite understand what you mean by "static copy".

Can you let me know the steps so that I could understand it better and employ it myself if certain situations require it.


Thanks in advance,
Senthil Kumar .A

[Bill Hassell]

IODC is similar to an PC BIOS. The error is probably due to the starting and/or ending address of lvol2 and not the size. The K580 is really old and for years, there was a limitation of the physical addresses for boot disk elements. How large is the volume (lvol1?) in front of lvol2? This is a boot ROM limitation (IODC) that must be addressed for legacy machines like the K580.

[RAC_1]

For James.

I doing it in correct order. I am doing it in following order.

lvlnboot -b /dev/vgroot/lvol1
lvlnboot -r /dev/vgroot/lvol3
lvlnboot -s /dev/vgroot/lvol2
lvlnboot -d /dev/vgroot/lvol2
lvlnboot -R /dev/vgroot

For Senthil,

I am preparing the third copy of boot disk. (In addition to primary boot disk and it's mirror) I am following this procedure.

How to create a copy of system disk?

.1 Locate the disk that is going to be used as a system copy
ioscan -kfnC disk

.2 Initialize the disk as a boot disk
pvcreate -f -B /dev/rdsk/c#t#d#

.3 Create the directory under /dev
mkdir /dev/vgroot

.4 Check for already exists minor numbers
ll /dev/vg*/group

.5 Create special file group from under /dev/vgroot using a unic minor number
mknod /dev/vgroot/group c 64 0x0#0000

.6 Create the new volume group
vgcreate /dev/vgroot /dev/dsk/c#t#d#

.7 Install the boot program on the specified device file using mkboot
mkboot /dev/rdsk/c#t#d#

.8 Update the specified device file for autofile
mkboot -a "hpux (;0)/stand/vmunix" /dev/rdsk/c#t#d#


.9 Create the logical volumes with the same size of the original system disk (usually vg00)
lvcreate -C y -r n -L [Size in Mbytes] -n lvol1 /dev/vgroot
lvcreate -C y -r n -L [Size in Mbytes] -n lvol2 /dev/vgroot
lvcreate -C y -r n -L [Size in Mbytes] -n lvol3 /dev/vgroot
-C contiguous allocation ploicy is set to YES
-r bad block relocation policy

.10 Create the other 5 filesystems without -C and -r
lvcreate -L [Size in Mbytes] -n lvol4/5/6/7/8 /dev/vgroot

.11 Prepare LVM logical volume to be root , boot , swap and dump area
lvlnboot -v - verbose all the exists boot disk.
lvlnboot -b /dev/vgroot/lvol1
lvlnboot -r /dev/vgroot/lvol3
lvlnboot -s /dev/vgroot/lvol2
lvlnboot -d /dev/vgroot/lvol2
lvlnboot -R /dev/vgroot
The order of the commands MUST be according to this list

.12 Construct a new file system using newfs
newfs -F hfs /dev/vgroot/rlvol1 - /stand files system must be hfs
newfs -F vxfs /dev/vgroot/rlvol3/4/5/6/7/8 - all the other files system should be vxfs
* /dev/vgroot/lvol2 will use for swap therefor no newfs should be used on it.

.13 Create directories for mount the vgroot logical volumes
mkdir /rootvg /standrootvg /usrrootvg /optrootvg /localrootvg /varrootvg

.14 Mount the logical volumes of vgroot to the same spot as they are in vg00
mount /dev/vgroot/lvol1 /standrootvg
mount /dev/vgroot/lvol3 /rootvg
mount /dev/vgroot/lvol4 /localrootvg
mount /dev/vgroot/lvol5 /optrootvg
mount /dev/vgroot/lvol7 /usrrootvg
mount /dev/vgroot/lvol8 /varrootvg
Be carefully when mounting the files system make double check that the new mount points in vgroot are like vg00. From one system to another there might be some difference.

.15 Copy all the files system (except for tmp, crash and swap) using find & cpio
cd / ; find . -xdev | cpio -pxlmd /rootvg
cd /stand ; find . -xdev | cpio -pxlmd /standrootvg
cd /usr ; find . -xdev | cpio -pxlmd /usrrootvg
cd /opt ; find . -xdev | cpio -pxlmd /optrootvg
cd /local ; find . -xdev | cpio -pxlmd /localrootvg
cd /var ; find . -xdev | cpio -pxlmd /varrootvg

.16 Umount all the files system

.17 Check the files system using fsck to be clean
fsck -F hfs /dev/vgroot/lvol1
fsck -F vxfs /dev/vgroot/lvol3/4/5/6/7/8
fsck should not be perform on lvol2 - swap.

How to boot the machine using the copy disk?

.1 Since vg00 is no configured with the device file of the system disk copied we must first reconfigure it. To do that we must be in maintenance mode.

.2 Boot the machines and stop it in the pdc section.

.3 Search for the disks and boot from the new disk using isl/ipl.
Main Menu>Boot [boot path] isl

.4 In the isl boot into maintenance mode
ISL>hpux -lm

.5 In maintenance mode the LVM is not in used and we can reconfigure vg00

.6 Add the /sbin the PATH variable
export PATH=/sbin:$PATH

.7 Export the exists vg00
vgexport vg00

.8 Create the vg00 directory under /dev
mkdir /dev/vg00

.9 Create the special file group under vg00
mknod /dev/vg00/group c 64 0x000000

.10 Import the disk with vg00
vgimport /dev/vg00 /dev/dsk/c#t#d#

.11 Active the volume group
vgchange -a y vg00

.12 Prepare LVM logical volume to be root , boot , swap and dump area but first remove the exists configuration
lvrmboot -r /dev/vg00
lvlnboot -v
lvlnboot -b /dev/vg00/lvol1
lvlnboot -r /dev/vg00/lvol3
lvlnboot -s /dev/vg00/lvol2
lvlnboot -d /dev/vg00/lvol2
lvlnboot -R

.13 Check files system using fsck to make sure they are clean
fsck -F hfs /dev/vg00/lvol1
fsck -F vxfs /dev/vg00/lvol3-8
make sure not to fsck lvol2 - swap

.14 Set the primary boot path to the new hardware path
setboot -p #/#/#.#.#

.15 Now we can boot into multi user mode by executing init 3 or reboot the machine.


For Bill,

I understand this is old system, but fact is that vg00 has everything right. Check my lvlnboot outout posted.

The lvol sizes for vgroot (for first three lvols is as follows.

LV Name /dev/vgroot/lvol1
LV Size (Mbytes) 300
LV Name /dev/vgroot/lvol2
LV Size (Mbytes) 4096
LV Name /dev/vgroot/lvol3
LV Size (Mbytes) 300

I created them as follows.
lvcreate -C y -r n options.

[Kent Ostby]

Okay. Bill's explanation got me digging.

This HP document explains part of the issue:
KBRC00002485

11.11 should have patches to handle it I would think. I know this was orginally patched as far back as 10.20.

[RAC_1]

Kent,

Can you please post link to document.
I am unable to get KBRC00002485

[Senthil Kumar .A_1]

Thank you very much RAC.

[Bill Hassell]

The KBRC00002485 document may require a software support contract to obtain. What the document describes is a mirroring situation where the primary boots fine but the mirror copy does not because the physical order of the lvols on the mirror pushed the primary swap area beyond 2Gb (or 4Gb depending on the CPU's ROMs). Make your sure that your lvol1 is small, or that the combination of lvol1+lvol2 does not extend beyond 4Gb (2Gb for older hardware).

[RAC_1]

Bill,

I saw the document. As you are mentioning, I am trying to understand, how it is already there on vg00? At the time of preparing vg00, we had no problems. The sizes of vg00 are same that I use for vgroot preparation. (Check vgroot sizes I posted). So whay it should happen only with vgroot and did not with vgoo? Will it matter (As Kent suggested) the firmware I/O of i/o cards?

Kent,

Any suggestions?

[RAC_1]

Bill,

vg00-primary disk pvdisplay -v
/dev/vg00/lvol1 75 75
/dev/vg00/lvol2 1024 1024
/dev/vg00/lvol3 75 75
/dev/vg00/lvol4 16 16
/dev/vg00/lvol5 320 320
/dev/vg00/lvol6 128 128
/dev/vg00/lvol7 256 256

vgroot-primary disk pvdisplay -v
/dev/vgroot/lvol1 75 75
/dev/vgroot/lvol2 1024 1024
/dev/vgroot/lvol3 75 75
/dev/vgroot/lvol4 32 32
/dev/vgroot/lvol5 640 640
/dev/vgroot/lvol6 256 256
/dev/vgroot/lvol7 67 67


Exactly same..

[Bill Hassell]

One last check:

lvdisplay -v /dev/vgroot/lvol1 | head
lvdisplay -v /dev/vgroot/lvol1 | head

This will show the actual extent assignment for lvol1 and lvol2. lvol1 should start at extent 1 and lvol2 should start at the end of lvol2, both pointing to the 3rd disk.

[RAC_1]

lvdisplay -v /dev/vgroot/lvol1
--- Logical volumes ---
LV Name /dev/vgroot/lvol1
VG Name /dev/vgroot
LV Permission read/write
LV Status available/syncd
Mirror copies 0
Consistency Recovery MWC
Schedule parallel
LV Size (Mbytes) 300
Current LE 75
Allocated PE 75
Stripes 0
Stripe Size (Kbytes) 0
Bad block off
Allocation strict/contiguous
IO Timeout (Seconds) default

--- Distribution of logical volume ---
PV Name LE on PV PE on PV
/dev/dsk/c5t5d0 75 75

--- Logical extents ---
LE PV1 PE1 Status 1
00000 /dev/dsk/c5t5d0 00000 current
00001 /dev/dsk/c5t5d0 00001 current

Some more goes here. Last line of above output.
00073 /dev/dsk/c5t5d0 00073 current
00074 /dev/dsk/c5t5d0 00074 current

lvdisplay -v /dev/vgroot/lvol2
--- Logical volumes ---
LV Name /dev/vgroot/lvol2
VG Name /dev/vgroot
LV Permission read/write
LV Status available/syncd
Mirror copies 0
Consistency Recovery MWC
Schedule parallel
LV Size (Mbytes) 4096
Current LE 1024
Allocated PE 1024
Stripes 0
Stripe Size (Kbytes) 0
Bad block off
Allocation strict/contiguous
IO Timeout (Seconds) default

--- Distribution of logical volume ---
PV Name LE on PV PE on PV
/dev/dsk/c5t5d0 1024 1024

--- Logical extents ---
LE PV1 PE1 Status 1
00000 /dev/dsk/c5t5d0 00075 current
00001 /dev/dsk/c5t5d0 00076 current
some more data here. Last lines as follows.
01022 /dev/dsk/c5t5d0 01097 current
01023 /dev/dsk/c5t5d0 01098 current


Also, would be possible to check i/o card iodc version? With STM?

[RAC_1]

Pushing it to top to get some more replies. I am still having this problem.

[RAC_1]

Firmware on all disks is same.

/dev/dsk/c5t5d0 - 10/4/4.5.0
Device Type: SCSI Disk Firmware Rev: HP06

/dev/dsk/c5t11d0 - 10/4/4.11.0
Device Type: SCSI Disk Firmware Rev: HP06

/dev/dsk/c4t6d0 - 10/0.6.0
Device Type: SCSI Disk Firmware Rev: HP06

/dev/dsk/c4t10d0 - 10/0.10.0
Device Type: SCSI Disk Firmware Rev: HP06

[Sameer_Nirmal]

RAC,

The problem being faced is indeed on account of different disk bus/interace used for the disk drive. The difference could be noted looking at the "ioscan" outut posted.
For vg00 disks the "sdisk" driver is used having differnt bus/interface nexus than vgroot disk which is using "disc3" driver.

The swap lvol seems to be falling beyond the maximum addressable range supported by the IODC of disk interface.

I would run CSTM Info-tool and Expert tool to get hardware details including IODC version. Looking at "ioscan -fn" output for
hardware at and below 10/4 would be useful too.

[RAC_1]

Sameer,

I also noticed that. Eeven driver for scsi interface is different. scsi1 and scsi2.
Also as you informed disk driver is also different. How that would happen?

What would cause it to use different driver?

[Torsten. · Accepted solution]

Hi RAC,

you have different drivers, because you have different BUS structures and different controllers.

10/4/4 is bus type HP PB slot 3

10/0 is bus type HP HSC on the core IO - internal and external drives

You can see the same for tape drives, some are claimed by stape, some by tape2 - depends also on the type of BUS and controller.