lvm -- lvm_chr_ioctl: unknown command 0x8004fe97

Since i want to run with ext3 and use snapshots, i applied the

	linux-2.4.18-VFS-lock.patch

to my default 2.4.18 kernel. I just did that file since making the patches, and applying that would fail during compilation. (something about md driver ?, which i dont use)

So, after just getting the VFS-lock patch to work, i've tried to use it on my debian woody, with

	lvm-common     1.5.5
	lvm10          1.0.4-4
installed. But i get alot of errors :/
	lvm -- lvm_chr_ioctl: unknown command 0x8004fe97
	lvm -- lvm_chr_ioctl: unknown command 0x8004fe97
	lvm -- lvm_chr_ioctl: unknown command 0x8004fe97
	lvm -- lvm_chr_ioctl: unknown command 0x8004fe97

And sometimes vgscan cant find my VG, and i cant activate it with
vgchange -a y ...

It _MIGHT_ be because i'm running lvm ontop of a loop-AES, but it does work most of the time.

JonB
ps: right now i'm starting all over with

	2.4.19
	lvm_1.1-rc2.tar.gz

and removing those debian packages.

You really don't want to do that... 1.1 is well broken.

Quoted 1 line

and removing those debian packages.

The Debian packages should be fine. Plenty of other people are using them.

Can you tell which command is issued the illegal ioctl ? It could be that the LVM device has a wrong major/minor or something else has take the LVM major/minor device number, because those are /not/ LVM ioctl numbers.

If you can determine what command is issuing those odd ioctls then you're probably half-way to finding the culprit.

Patrick Caulfield wrote:

[cut]

Quoted 1 line

You really don't want to do that... 1.1 is well broken.

uhoh, i'm already doing it, not production yet, but soon. I figured that the latest from oktober 3, 1.1-rc2 would be good enough.

Quoted 2 lines

Quoted 1 line

and removing those debian packages.

Quoted 5 lines

The Debian packages should be fine. Plenty of other people are using them.

Can you tell which command is issued the illegal ioctl ? It could be that the LVM device has a wrong major/minor or something else has take the LVM major/minor device number, because those are /not/ LVM ioctl numbers.

More than one. vgchange -a y <VG>
lvremove (snapshopts)

Quoted 1 line · Jon Bendtsen

Patrick Caulfield wrote:

Quoted 2 lines

More than one. vgchange -a y <VG>
lvremove (snapshopts)

It does sound a lot like an "odd" lvm toolset (or perhaps library) that's interfering. It's worth checking whicg vgchange (etc) command is being run (ie is there one in /usr/local or /usr/sbin that's earlier in the PATH than the "real" one. Also do "ldd" in the library to see which library is being picked up.

Quoted 3 lines · Patrick Caulfield

Quoted 1 line · Jon Bendtsen

Patrick Caulfield wrote:

[cut]

Quoted 8 lines

Quoted 2 lines

More than one. vgchange -a y <VG>
lvremove (snapshopts)

It does sound a lot like an "odd" lvm toolset (or perhaps library) that's interfering. It's worth checking whicg vgchange (etc) command is being run (ie is there one in /usr/local or /usr/sbin that's earlier in the PATH than the "real" one. Also do "ldd" in the library to see which library is being picked up.

i think it was the debian packages, because it stopped after removing the
debian packages and make;make install
the lvm-1.1-rc2

On Tue, Oct 22, 2002 at 02:57:29PM +0200, Jon Bendtsen wrote:

Quoted 4 lines

i think it was the debian packages, because it stopped after removing the
debian packages and make;make install
the lvm-1.1-rc2

Seems unlikely - I'm the maintainer of the Debian packages and I can state categorically that they don't issue that ioctl :-)

Quoted 2 lines · Patrick Caulfield

On Tue, Oct 22, 2002 at 02:57:29PM +0200, Jon Bendtsen wrote:

[cut]

Quoted 5 lines

Quoted 4 lines

i think it was the debian packages, because it stopped after removing the
debian packages and make;make install
the lvm-1.1-rc2

Quoted 2 lines

Seems unlikely - I'm the maintainer of the Debian packages and I can state categorically that they don't issue that ioctl :-)

well, it works now. I did try with another lvm 1.1-rc? maybe thats why?