[linux-lvm] disk add means trouble: vgscan -- invalid i/o protocol version 5

[linux-lvm] disk add means trouble: vgscan -- invalid i/o protocol version 5

[Stefan Helfert]

Dear list,

I was running a stable configuration until yesterday:
several SCSI disks with standard partitions, two IDE disks, only one IDE
disk (Maxtor 40Gig /dev/hda) holding a LVM partition. LVM version that
of 
SuSE 7.0, that was lvm-0.8 Build 111 from SuSE.

When I added a 60 GB Maxtor IDE as primary slave, I found that vgscan
only
throws an "invalid i/o protocol" error. Upgrading to a SuSE 7.0 update
lvm did not help. Upgrading to the SuSE 7.1 lvm-0.9.1_beta-3 did not
help
except detailing the error "... protocol version 5".

When detaching the 60 GB disk, everything works again. (The 60 Gig disk 
is empty, it did not hold a partition table -- and it does not work with
a
partition table and some partition either.)

Any ideas ? (The 40Gig disk is in dire need of replacement, and this
configuration should be used over the weekend to move the data).

Let me stress this:
the configuration had been working flawlessly before, and I checked that
/dev/lvm is around.


Stefan

bash-2.04# vgscan -d                                                   
<1> lvm_get_iop_version -- CALLED
<22> lvm_check_special -- CALLED
<22> lvm_check_special -- LEAVING
<1> lvm_get_iop_version -- AFTER ioctl ret: 0 
<1> lvm_get_iop_version -- LEAVING with ret: 5   
vgscan -- invalid i/o protocol version 5

Re: [linux-lvm] disk add means trouble: vgscan -- invalid i/o protocol version 5

[AJ Lewis]

[-- Attachment #1: Type: text/plain, Size: 2564 bytes --]

On Mon, Apr 16, 2001 at 08:34:18PM +0200, Stefan Helfert wrote:
> Dear list,
> 
> I was running a stable configuration until yesterday:
> several SCSI disks with standard partitions, two IDE disks, only one IDE
> disk (Maxtor 40Gig /dev/hda) holding a LVM partition. LVM version that
> of 
> SuSE 7.0, that was lvm-0.8 Build 111 from SuSE.
> 
> When I added a 60 GB Maxtor IDE as primary slave, I found that vgscan
> only
> throws an "invalid i/o protocol" error. Upgrading to a SuSE 7.0 update
> lvm did not help. Upgrading to the SuSE 7.1 lvm-0.9.1_beta-3 did not
> help
> except detailing the error "... protocol version 5".
> 
> When detaching the 60 GB disk, everything works again. (The 60 Gig disk 
> is empty, it did not hold a partition table -- and it does not work with
> a
> partition table and some partition either.)

Can you run vgscan w/o the 60GB disk attached?  If so, that's very
odd...there may be some crud on the front of the disk that the LVM tools are
reading as metadata.

If you've updated the tools, you need to upgrade the kernel as well.  That's
normally why you'd get the invalid IOP protocol message.

> Any ideas ? (The 40Gig disk is in dire need of replacement, and this
> configuration should be used over the weekend to move the data).
> 
> Let me stress this:
> the configuration had been working flawlessly before, and I checked that
> /dev/lvm is around.
> 
> 
> Stefan
> 
> bash-2.04# vgscan -d                                                   
> <1> lvm_get_iop_version -- CALLED
> <22> lvm_check_special -- CALLED
> <22> lvm_check_special -- LEAVING
> <1> lvm_get_iop_version -- AFTER ioctl ret: 0 
> <1> lvm_get_iop_version -- LEAVING with ret: 5   
> vgscan -- invalid i/o protocol version 5
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm

-- 
AJ Lewis
Sistina Software Inc.                  Voice:  612-379-3951
1313 5th St SE, Suite 111              Fax:    612-379-3952
Minneapolis, MN 55414                  E-Mail: lewis@sistina.com
http://www.sistina.com

Current GPG fingerprint = 3B5F 6011 5216 76A5 2F6B  52A0 941E 1261 0029 2648
Get my key at: http://www.sistina.com/~lewis/gpgkey
 (Unfortunately, the PKS-type keyservers do not work with multiple sub-keys)

-----Begin Obligatory Humorous Quote----------------------------------------
Hey! It compiles! Ship it!
-----End Obligatory Humorous Quote------------------------------------------

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: [linux-lvm] disk add means trouble: vgscan -- invalid i/o protocol version 5

[Stefan Helfert]

AJ Lewis wrote:
> 
> On Mon, Apr 16, 2001 at 08:34:18PM +0200, Stefan Helfert wrote:
> > Dear list,
> >
> > I was running a stable configuration until yesterday:
> > several SCSI disks with standard partitions, two IDE disks, only one IDE
> > disk (Maxtor 40Gig /dev/hda) holding a LVM partition. LVM version that
> > of
> > SuSE 7.0, that was lvm-0.8 Build 111 from SuSE.
> >
> > When I added a 60 GB Maxtor IDE as primary slave, I found that vgscan
> > only
> > throws an "invalid i/o protocol" error. Upgrading to a SuSE 7.0 update
> > lvm did not help. Upgrading to the SuSE 7.1 lvm-0.9.1_beta-3 did not
> > help
> > except detailing the error "... protocol version 5".
> >
> > When detaching the 60 GB disk, everything works again. (The 60 Gig disk
> > is empty, it did not hold a partition table -- and it does not work with
> > a
> > partition table and some partition either.)
> 
> Can you run vgscan w/o the 60GB disk attached?  If so, that's very
> odd...there may be some crud on the front of the disk that the LVM tools are
> reading as metadata.

I'm sorry to tell:
The 40 Gig disk alone, w/o 60GB disk attached, vgscan runs.
I now have moved the data with the help of another machine, and
currently the 40 GB disk is primary master and the 60GB disk is primary
slave -- and vgscan does not run anymore. Note that the 60GB disk is
holding partitions now -- a NTFS partition and a ext2, but no LVM
partition there.

Ideas how to investigate ?

> 
> If you've updated the tools, you need to upgrade the kernel as well.  That's
> normally why you'd get the invalid IOP protocol message.

I understand. I did downgrade to LVM-0.8 again, since 0.9.1-beta3 did
not run with my 2.2 kernel.


Stefan

Re: [linux-lvm] disk add means trouble: vgscan -- invalid i/o protocol version 5

[Stefan Helfert]

I am reposting this, since I believe I have run into a serious condition
here:

A stable, running configuration consisted of one IDE disk with one LVM
VG on it, and adding another IDE disk leads to a non-working vgscan. If
I go back to the one IDE disk configuration, the LVM gets discovered,
and adding the second disk again leads to non-working vgscan. 
I cannot completely rule out any hardware issues here (since the disk
with the LVM VG is apparently having lots of SMART redirects and
significant access time increases), but I'd like to investigate this
vgscan problem further.
(Since it reduces my confidence in LVM very much, so much that I am
currently killing the LVM on my terabyte raid because I do not want to
see anything like that there.)

So if anyone has an idea of how to investigate this, I am open to
experiments. If not, I'll send the disk back to the manufacturer for
replacement.

Stefan

AJ Lewis wrote:
> 
> On Mon, Apr 16, 2001 at 08:34:18PM +0200, Stefan Helfert wrote:
> > Dear list,
> >
> > I was running a stable configuration until yesterday:
> > several SCSI disks with standard partitions, two IDE disks, only one IDE
> > disk (Maxtor 40Gig /dev/hda) holding a LVM partition. LVM version that
> > of
> > SuSE 7.0, that was lvm-0.8 Build 111 from SuSE.
> >
> > When I added a 60 GB Maxtor IDE as primary slave, I found that vgscan
> > only
> > throws an "invalid i/o protocol" error. Upgrading to a SuSE 7.0 update
> > lvm did not help. Upgrading to the SuSE 7.1 lvm-0.9.1_beta-3 did not
> > help
> > except detailing the error "... protocol version 5".
> >
> > When detaching the 60 GB disk, everything works again. (The 60 Gig disk
> > is empty, it did not hold a partition table -- and it does not work with
> > a
> > partition table and some partition either.)
> 
> Can you run vgscan w/o the 60GB disk attached?  If so, that's very
> odd...there may be some crud on the front of the disk that the LVM tools are
> reading as metadata.

I'm sorry to tell:
The 40 Gig disk alone, w/o 60GB disk attached, vgscan runs.
I now have moved the data with the help of another machine, and
currently the 40 GB disk is primary master and the 60GB disk is primary
slave -- and vgscan does not run anymore. Note that the 60GB disk is
holding partitions now -- a NTFS partition and a ext2, but no LVM
partition there.

Ideas how to investigate ?

> 
> If you've updated the tools, you need to upgrade the kernel as well.  That's
> normally why you'd get the invalid IOP protocol message.

I understand. I did downgrade to LVM-0.8 again, since 0.9.1-beta3 did
not run with my 2.2 kernel.


Stefan
_______________________________________________
linux-lvm mailing list
linux-lvm@sistina.com
http://lists.sistina.com/mailman/listinfo/linux-lvm

Re: [linux-lvm] disk add means trouble: vgscan -- invalid i/o protocol version 5

[Patrick Caulfield]

On Fri, Apr 20, 2001 at 01:28:42AM +0200, Stefan Helfert wrote:
> I am reposting this, since I believe I have run into a serious condition
> here:
> 
> A stable, running configuration consisted of one IDE disk with one LVM
> VG on it, and adding another IDE disk leads to a non-working vgscan. If
> I go back to the one IDE disk configuration, the LVM gets discovered,
> and adding the second disk again leads to non-working vgscan. 
> I cannot completely rule out any hardware issues here (since the disk
> with the LVM VG is apparently having lots of SMART redirects and
> significant access time increases), but I'd like to investigate this
> vgscan problem further.
> (Since it reduces my confidence in LVM very much, so much that I am
> currently killing the LVM on my terabyte raid because I do not want to
> see anything like that there.)
> 
> So if anyone has an idea of how to investigate this, I am open to
> experiments. If not, I'll send the disk back to the manufacturer for
> replacement.


The first thing to do is to get your kernel and user tools in sync so that 
you don't get th IOP message. If you have a kernel you like then node the LVM
version number in that kernel and build the tools for that version.

That should get vgscan working after some fashion. If the new disk still seems
to be causing the problem then please post the error here so we can have a look
at it.


patrick

Re: [linux-lvm] disk add means trouble: vgscan -- invalid i/o protocol version 5

[Stefan Helfert]

Patrick Caulfield wrote:
> 
> On Fri, Apr 20, 2001 at 01:28:42AM +0200, Stefan Helfert wrote:
> > I am reposting this, since I believe I have run into a serious condition
> > here:
> >
> > A stable, running configuration consisted of one IDE disk with one LVM
> > VG on it, and adding another IDE disk leads to a non-working vgscan. If
> > I go back to the one IDE disk configuration, the LVM gets discovered,
> > and adding the second disk again leads to non-working vgscan.
> > I cannot completely rule out any hardware issues here (since the disk
> > with the LVM VG is apparently having lots of SMART redirects and
> > significant access time increases), but I'd like to investigate this
> > vgscan problem further.
> > (Since it reduces my confidence in LVM very much, so much that I am
> > currently killing the LVM on my terabyte raid because I do not want to
> > see anything like that there.)
> >
> > So if anyone has an idea of how to investigate this, I am open to
> > experiments. If not, I'll send the disk back to the manufacturer for
> > replacement.
> 
> The first thing to do is to get your kernel and user tools in sync so that
> you don't get th IOP message. If you have a kernel you like then node the LVM
> version number in that kernel and build the tools for that version.
> 
> That should get vgscan working after some fashion. If the new disk still seems
> to be causing the problem then please post the error here so we can have a look
> at it.
> 
> patrick

My kernel and user tools are in sync (lvm_0.8) and my initial problem
here is that I get the IOP message -- and I get this message out of the
blue. No software changes, no kernel changes, nothing else except adding
another disk drive, the configuration had been running stable since
ages. I do not see anything here except the vgscan error, and the output
of "vgscan -d" is here:

bash-2.04# vgscan -d                                                   
<1> lvm_get_iop_version -- CALLED
<22> lvm_check_special -- CALLED
<22> lvm_check_special -- LEAVING
<1> lvm_get_iop_version -- AFTER ioctl ret: 0 
<1> lvm_get_iop_version -- LEAVING with ret: 5   
vgscan -- invalid i/o protocol version 5

??

Stefan

Re: [linux-lvm] disk add means trouble: vgscan -- invalid i/o protocol version 5

[Patrick Caulfield]

On Tue, Apr 24, 2001 at 08:53:03PM +0200, Stefan Helfert wrote:
> Patrick Caulfield wrote:
> 
> My kernel and user tools are in sync (lvm_0.8) and my initial problem
> here is that I get the IOP message -- and I get this message out of the
> blue. No software changes, no kernel changes, nothing else except adding
> another disk drive, the configuration had been running stable since
> ages. I do not see anything here except the vgscan error, and the output
> of "vgscan -d" is here:
> 
> bash-2.04# vgscan -d                                                   
> <1> lvm_get_iop_version -- CALLED
> <22> lvm_check_special -- CALLED
> <22> lvm_check_special -- LEAVING
> <1> lvm_get_iop_version -- AFTER ioctl ret: 0 
> <1> lvm_get_iop_version -- LEAVING with ret: 5   
> vgscan -- invalid i/o protocol version 5

Sorry, but if you're getting an invalid IOP message then the kernel and tools
can't possibly be in sync, that's what "in sync" means in this context.

There's nothing that can happen just by adding a disk drive (unless it has some
other tools on it that appear in the PATH) that can cause this.

All I can suggest is that you download the user tools that match your kernel and
rebuild them.

patrick