On Sat, 2006-07-08 at 19:48 +0300, Itay wrote:

From: William L. Maltby BillsCentOS@triad.rr.com

...

From: Itay itayf@nospammail.net

...

...

Also, I just started working with lvm. The following warnings are genertated whenever I run one of lvm commands on the external usb disk (MyBook by Western Digital).

[snip]

...

...

[root@frodo pilpel10]# vgcreate VGe1 /dev/sdc2

...

Did you do pvcreate?

As far as I remember -- yes. When I return to my office, at the beginning of next week, I'll look at the terminal session to see the exact command sequence. But my conviction is reinforced by the output of pvscan and pvdisplay (wasn't in the original message) below.

...

...

/dev/dm-11: read failed after 0 of 4096 at 5242814464: Input/output error /dev/dm-11: read failed after 0 of 4096 at 0: Input/output error /dev/dm-12: read failed after 0 of 4096 at 54460416: Input/output error /dev/dm-12: read failed after 0 of 4096 at 0: Input/output error

[snip]

...

...

Volume group "VGe1" successfully created

...

...

[root@frodo pilpel10]# pvscan /dev/dm-11: read failed after 0 of 4096 at 5242814464: Input/output error /dev/dm-11: read failed after 0 of 4096 at 0: Input/output error

[snip]

...

...

PV /dev/sdc2 VG VGe1 lvm2 [296.13 GB / 296.13 GB free] PV /dev/sda2 VG VG00 lvm2 [232.78 GB / 1.78 GB free] Total: 2 [528.91 GB] / in use: 2 [528.91 GB] / in no VG: 0 [0 ]

[root@frodo ~]# pvdisplay /dev/sdc2 /dev/dm-11: read failed after 0 of 4096 at 0: Input/output error /dev/dm-12: read failed after 0 of 4096 at 0: Input/output error /dev/dm-13: read failed after 0 of 4096 at 0: Input/output error /dev/dm-14: read failed after 0 of 4096 at 0: Input/output error --- Physical volume --- PV Name /dev/sdc2 VG Name VGe1 PV Size 296.13 GB / not usable 0 Allocatable yes PE Size (KByte) 4096 Total PE 75810 Free PE 24098 Allocated PE 51712 PV UUID Ue56s5-nkax-csIU-6iBX-l4eb-hIJ1-nxJLgl

Thanks, Itay

We are about to exceed my knowledge, as I have never used USB for anything other than thumb drives on MS Win*. I believe I asked about partitioning? If you do

sfdisk -l /dev/sdc

what does it show? If you do a

dd if=/dev/sdc2 of=/dev/null

do you get I/O errors? If so, my guess is it's not an LVM problem.

---

CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos

On Sat, 2006-07-08 at 21:01 +0300, Itay wrote:

On Sat, 2006-07-08 at 19:48 +0300, Itay wrote:

...

From: William L. Maltby BillsCentOS@triad.rr.com

[snip]

...

...

<snip>

...

We are about to exceed my knowledge, as I have never used USB for anything other than thumb drives on MS Win*. I believe I asked about partitioning? If you do

sfdisk -l /dev/sdc

what does it show? If you do a

dd if=/dev/sdc2 of=/dev/null

[root@frodo ~]# sfdisk -l /dev/sdc

Disk /dev/sdc: 38913 cylinders, 255 heads, 63 sectors/track Units = cylinders of 8225280 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start     End   #cyls    #blocks   Id  System

/dev/sdc1 * 0+ 254 255- 2048256 c W95 FAT32 (LBA) /dev/sdc2 255 38912 38658 310520385 83 Linux /dev/sdc3 0 - 0 0 0 Empty /dev/sdc4 0 - 0 0 0 Empty

** I note that the file system on /dev/sdc2 is 'Linux' and not 'Linux LVM'. This is what I got, eventhough I created it with parted /dev/sdc mkpart ...

There's your problem I think. In my first post, I *guessed* that type needed to be 86. I'm sure 83 is NG - that's a standard Linux part. With fdisk and it's variations, there is a command to change partition type. Do that and select the one that shows LVM 986 I think). Write and (possibly) reboot or sfdisk -R /dev/sdc (which re-reads the partition info. If it shows *LVM* as the type, do you pvcreate, vgcreate, etc. and I bet you are OK.

[root@frodo ~]# dd if=/dev/sdc2 of=/dev/null 12747056+0 records in 12747056+0 records out

** The above output is after letting dd running for few minutes; after that I killed the process because I was afraid it will take it too long.

Yes. The purpose was to see that partition size is OK (doesn't run off end of device). If you add blksize= with 4096, 8192 or 16384 or ... it goes much faster. But I think no need now, it's your partition type I think.

Thank you for your help,

Us n00bs have to stick together here! :-)

Itay

<snip sig stuff>

HTH

On Sat, 2006-07-08 at 14:13 -0400, William L. Maltby wrote:

On Sat, 2006-07-08 at 21:01 +0300, Itay wrote:

...

On Sat, 2006-07-08 at 19:48 +0300, Itay wrote:

...

From: William L. Maltby BillsCentOS@triad.rr.com

<snip>

Us n00bs have to stick together here! :-)

And being n00bs, we are permitted to expound at length with guesses about what might be going on, giving the wizened cranky old pros someone to pick on and soothe their savage egos! ;-) I've not read any devel docs yet, so there may be a lot more going on.

From rom my reading, I think the device mapper (?) needs to see a partition 86 to know to look inside the partition for the meta-data (placed there by pvcreate/lvcreate?) that is used to generate the devices. My *guess* is if no type 86 is found it means no pvdata is garnered and no /dev/mapper/VolGroup* and no /dev/VolGroup* symlinks are made (or wrong ones are made or left over from elsewhere?).

I think if it finds one or more type 86 partitions, it builds all the /dev/mapper lvms and symlinks it needs. This includes proper association of minor device to vg so you don have multiple physical vols switching places on you every time you boot.

Anyway, that's my SWAG (Scientific Wild-Assed Guess) about what's going on.

<snip sig stuff>

When it's all working, be sure and post with a "[SOLVED]" added to the subject and a summary of your fixes as a courtesy to other list members and future searchers.

On Sat, 2006-07-08 at 14:38 -0400, Matt Hyclak wrote:

On Sat, Jul 08, 2006 at 02:35:51PM -0400, William L. Maltby enlightened us:

...

...

On Sat, 2006-07-08 at 21:01 +0300, Itay wrote:

...

On Sat, 2006-07-08 at 19:48 +0300, Itay wrote:

...

From: William L. Maltby BillsCentOS@triad.rr.com <snip><snip>

Type 86 is "NTFS volume set". I think you mean 8e which is "Linux LVM".

Thanks! I knew I wasn't sure of the type. I don't know why 86 stuck in my head.

Matt

...

[root@frodo ~]# sfdisk -l /dev/sdc

Disk /dev/sdc: 38913 cylinders, 255 heads, 63 sectors/track Units = cylinders of 8225280 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start     End   #cyls    #blocks   Id  System

/dev/sdc1 * 0+ 254 255- 2048256 c W95 FAT32 (LBA) /dev/sdc2 255 38912 38658 310520385 83 Linux /dev/sdc3 0 - 0 0 0 Empty /dev/sdc4 0 - 0 0 0 Empty

** I note that the file system on /dev/sdc2 is 'Linux' and not 'Linux LVM'. This is what I got, eventhough I created it with parted /dev/sdc mkpart ...

There's your problem I think. In my first post, I *guessed* that type needed to be 86. I'm sure 83 is NG - that's a standard Linux part. With fdisk and it's variations, there is a command to change partition type. Do that and select the one that shows LVM 986 I think). Write and (possibly) reboot or sfdisk -R /dev/sdc (which re-reads the partition info. If it shows *LVM* as the type, do you pvcreate, vgcreate, etc. and I bet you are OK.

Had to wait until I came back to work and could reboot the machine. Using William's instructions, modulus using 8e as Matt Hyclak pointed out, seems to solve the problem. Everything seems to work fine: lvcreation, writing files, etc.

The only odd thing is in regard to the output of sfdisk. Note the warning when 'sfdisk -l' is invoked without a device argument. According to the man page this is a legitimate call that should produce output similar in nature to 'fdisk -l' (which works as expected).

<man> The second type of invocation: sfdisk -l [options] device will list the partitions on this device. If the device argument is omitted, the partitions on all hard disks are listed. % sfdisk -l /dev/hdc </man>

I am quite sure that this is an 'sfdisk' issue unrelated to the problem I needed to solve. But I would like to be careful before I declare [RESOLVED].

Perhaps I will try sfdisk with a different machine to see if there is a difference.

Thanks, Itay

<transcript> [root@frodo ~]# sfdisk -l

Disk /dev/sda: 30401 cylinders, 255 heads, 63 sectors/track Units = cylinders of 8225280 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start End #cyls #blocks Id System /dev/sda1 * 0+ 12 13- 104391 83 Linux /dev/sda2 13 30400 30388 244091610 8e Linux LVM /dev/sda3 0 - 0 0 0 Empty /dev/sda4 0 - 0 0 0 Empty Warning: start=63 - this looks like a partition rather than the entire disk. Using fdisk on it is probably meaningless. [Use the --force option if you really want this]

[root@frodo ~]# sfdisk -l /dev/sda Disk /dev/sda: 30401 cylinders, 255 heads, 63 sectors/track Units = cylinders of 8225280 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start End #cyls #blocks Id System /dev/sda1 * 0+ 12 13- 104391 83 Linux /dev/sda2 13 30400 30388 244091610 8e Linux LVM /dev/sda3 0 - 0 0 0 Empty /dev/sda4 0 - 0 0 0 Empty

[root@frodo ~]# sfdisk -l /dev/sdb Disk /dev/sdb: 38913 cylinders, 255 heads, 63 sectors/track Units = cylinders of 8225280 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start End #cyls #blocks Id System /dev/sdb1 * 0+ 254 255- 2048256 c W95 FAT32 (LBA) /dev/sdb2 255 38912 38658 310520385 8e Linux LVM /dev/sdb3 0 - 0 0 0 Empty /dev/sdb4 0 - 0 0 0 Empty </transcript>

On Mon, 10 Jul 2006 11:13:42 -0400 William L. Maltby BillsCentOS@triad.rr.com wrote:

...

I am quite sure that this is an 'sfdisk' issue unrelated to the problem I needed to solve. But I would like to be careful before I declare [RESOLVED].

My workstation, in use for over a year now, shows the same warning message (start=63...) as your transcript. Further, having used sfdisk for a *long* time for some fairly rigorous applications, I have complete confidence in it. Reread the man sections on the "+" and "-" suffix and you get your first clue. The run sfdisk again and add "-uS" and I think you will be OK with it.

[snip]

...

Perhaps I will try sfdisk with a different machine to see if there is a difference.

I don't think you'll see any difference if the config is similar. I've attached my transcript below.

Indeed your example (below) confirms that this is an sfdisk issue, unrelated to my LVM problem which I now consider resolved.

Well, this is what I call 2 (or 3) at the price of one. Thanks for the extra education!

Itay

============================================================= [root@wlmlfs08 ~]# sfdisk -l

Disk /dev/hda: 12161 cylinders, 255 heads, 63 sectors/track Units = cylinders of 8225280 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start End #cyls #blocks Id System /dev/hda1 * 0+ 12 13- 104391 83 Linux /dev/hda2 13 2445 2433 19543072+ 8e Linux LVM /dev/hda3 2446 4876 2431 19527007+ 8e Linux LVM /dev/hda4 0 - 0 0 0 Empty Warning: start=63 - this looks like a partition rather than the entire disk. Using fdisk on it is probably meaningless. [Use the --force option if you really want this]

[root@wlmlfs08 ~]# sfdisk -l -uS

Disk /dev/hda: 12161 cylinders, 255 heads, 63 sectors/track Units = sectors of 512 bytes, counting from 0

Device Boot Start End #sectors Id System /dev/hda1 * 63 208844 208782 83 Linux /dev/hda2 208845 39294989 39086145 8e Linux LVM /dev/hda3 39294990 78349004 39054015 8e Linux LVM /dev/hda4 0 - 0 0 Empty Warning: start=63 - this looks like a partition rather than the entire disk. Using fdisk on it is probably meaningless. [Use the --force option if you really want this] [root@wlmlfs08 ~]#

============================================================================

itayf@nospammail.net spake the following on 7/10/2006 7:09 AM:

...

...

[root@frodo ~]# sfdisk -l /dev/sdc

Disk /dev/sdc: 38913 cylinders, 255 heads, 63 sectors/track Units = cylinders of 8225280 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start     End   #cyls    #blocks   Id  System

/dev/sdc1 * 0+ 254 255- 2048256 c W95 FAT32 (LBA) /dev/sdc2 255 38912 38658 310520385 83 Linux /dev/sdc3 0 - 0 0 0 Empty /dev/sdc4 0 - 0 0 0 Empty

** I note that the file system on /dev/sdc2 is 'Linux' and not 'Linux LVM'. This is what I got, eventhough I created it with parted /dev/sdc mkpart ...

There's your problem I think. In my first post, I *guessed* that type needed to be 86. I'm sure 83 is NG - that's a standard Linux part. With fdisk and it's variations, there is a command to change partition type. Do that and select the one that shows LVM 986 I think). Write and (possibly) reboot or sfdisk -R /dev/sdc (which re-reads the partition info. If it shows *LVM* as the type, do you pvcreate, vgcreate, etc. and I bet you are OK.

Had to wait until I came back to work and could reboot the machine. Using William's instructions, modulus using 8e as Matt Hyclak pointed out, seems to solve the problem. Everything seems to work fine: lvcreation, writing files, etc.

The only odd thing is in regard to the output of sfdisk. Note the warning when 'sfdisk -l' is invoked without a device argument. According to the man page this is a legitimate call that should produce output similar in nature to 'fdisk -l' (which works as expected).

<man> The second type of invocation: sfdisk -l [options] device will list the partitions on this device. If the device argument is omitted, the partitions on all hard disks are listed. % sfdisk -l /dev/hdc </man>

Since programmers are notorious for putting off the documentation, it is very common to have man pages not completely in sync with a program.

Scott Silva wrote:

Since programmers are notorious for putting off the documentation, it is very common to have man pages not completely in sync with a program.

That's one area where OpenBSD is far superior to Linux... the man pages are considered to be *the* authoritative documentation for any program. Any inaccuracies or incompleteness is considered to be a serious bug in the program.

Barry