RE: [linux-lvm] Bonehead problem

RE: [linux-lvm] Bonehead problem

[Verloove, Olivier]

In fact, I didn't have LVM 0.5 compiled on this system.
The problem was due to the device file /dev/lvm missing.

And thanks for the md[4-15] remark. I'm on RH5.2 and it was the cause of
vgscan hanging the system (see other message).

Olivier
> ----------
> From:
> Steve_J_Brueggeman@notes.seagate.com[SMTP:Steve_J_Brueggeman@notes.seagate
> .com]
> Sent: 	Thursday, 11 Mar, 1999 16:08
> To: 	linux-lvm@msede.com
> Subject: 	RE: [linux-lvm] Bonehead problem
> 
> 
> It sounds to me like you've got old modules hanging around as left-overs.
> Did you build LVM 0.5 with the same kernel version? (2.2.2 presumably).
> If
> so, did you clear out your old modules directory before you rebuilt the
> kernel with LVM 0.6?  Are you using modversions when you build the
> kernel/modules?
> 
> I've also noticed that vgscan will cause the kernel will complain about
> not
> being able to find some block devices that I have not compiled into
> modules, or the kernel.  This doesn't sound like your problem though.
> 
> I've also had a problem when vgscan ran, because of a bug in the md
> driver.
> I run RedHat 5.2, and the default installation created /dev/md0 through
> /dev/md15, but the default md driver is compiled to handle only 4 md
> devices (#define MAX_MD_DEV  4), and because of a bug in the md_open()
> (uses [minor] without checking minor is less than MAX_MD_DEV), my kernel
> would crash when vgscan tried to check /dev/md6 or so.  I just rm'd
> /dev/md[4-9] /dev/md1[0-5] to not cause the problem.   I don't think this
> is your problem either, but thought you might like to know.
> 
> Steve Brueggeman
> (All opinions are my own, and in no way reflect those of my employer).
> (<--legal-crud)
> 
> 
> 
> 
> 
> 

RE: [linux-lvm] Bonehead problem

[Verloove, Olivier]

Yes... indeed, creating the device file solved the problem.

Now, I ran vgscan and it hanged the system :-(
Here's what I found in the syslog:

Mar 11 16:07:17 cobalt modprobe: can't locate module char-major-109
Mar 11 16:07:20 cobalt kernel: LVM version 0.6  by Heinz Mauelshagen
(26/02/199
9)
Mar 11 16:07:20 cobalt kernel: lvm -- Module successfully initialized
Mar 11 16:07:24 cobalt kernel: cdrom: open failed.
Mar 11 16:07:24 cobalt modprobe: can't locate module block-major-22
Mar 11 16:07:24 cobalt last message repeated 5 times
Mar 11 16:07:24 cobalt modprobe: can't locate module block-major-33
Mar 11 16:07:24 cobalt last message repeated 5 times
Mar 11 16:07:24 cobalt modprobe: can't locate module block-major-34
Mar 11 16:07:24 cobalt last message repeated 5 times
Mar 11 16:07:24 cobalt kernel: VFS: grow_buffers: size = 16
Mar 11 16:07:25 cobalt last message repeated 2613 times
Mar 11 16:07:25 cobalt kernel: Kernel panic: Free list corrupted Mar 11
16:07:17 cobalt modprobe: can't locate module char-major-109
Mar 11 16:07:20 cobalt kernel: LVM version 0.6  by Heinz Mauelshagen
(26/02/199
9)
Mar 11 16:07:20 cobalt kernel: lvm -- Module successfully initialized
Mar 11 16:07:24 cobalt kernel: cdrom: open failed.
Mar 11 16:07:24 cobalt modprobe: can't locate module block-major-22
Mar 11 16:07:24 cobalt last message repeated 5 times
Mar 11 16:07:24 cobalt modprobe: can't locate module block-major-33
Mar 11 16:07:24 cobalt last message repeated 5 times
Mar 11 16:07:24 cobalt modprobe: can't locate module block-major-34
Mar 11 16:07:24 cobalt last message repeated 5 times
Mar 11 16:07:24 cobalt kernel: VFS: grow_buffers: size = 16
Mar 11 16:07:25 cobalt last message repeated 2613 times
Mar 11 16:07:25 cobalt kernel: Kernel panic: Free list corrupted 

I had a look in major.h and 22, 33, 34 corresponds to IDE1_MAJOR, IDE2_MAJOR
and IDE3_MAJOR ?

Any idea.

To summarize, this is a new install (no VG created) of Linux 2.2.2 (without
devfs) + LVM 0.6.
I have MD compiled in the kernel ... but no MD device created.

Thanks.

Olivier
> ----------
> From: 	Klaus Strebel[SMTP:stb@ep-ag.com]
> Reply To: 	antivirus@ep-ag.com
> Sent: 	Thursday, 11 Mar, 1999 15:56
> To: 	Verloove, Olivier
> Cc: 	linux-lvm@msede.com
> Subject: 	Re: [linux-lvm] Bonehead problem
> 
> Hi Olivier,
> 
> "Verloove, Olivier" wrote:
> > 
> > I have the same problem (Linux 2.2.2 + LVM 0.6).
> > With or without lvm module loaded, I get:
> >         vgscan -- invalid i/o protocol version
> > 
> > Is there another module which should be loaded at the same time ?
> > lsmod gives me:
> >         lvm     26960   0  (unused)
> Do you have the char-device /dev/lvm (Major 109, Minor 0) ? The LVM-0.5
> module
> seems to create this device, if it doesn't exist. The LVM-0.6 seems not to
> do
> this, i use devfs and i have to create the device each time before i load
> the
> module.
> 
> Cheers
> Klaus
> -- 
> Klaus Strebel
> stb@ep-ag.com
> EIGNER + PARTNER AG   - The Engineering Warehouse Company -
> <http://www.ep-ag.com>
> -----------------------------------------------------------------------
> In Heaven there will be no DOS, no Intel, and above all, no Bill Gates.
> 

RE: [linux-lvm] Bonehead problem

[Steve_J_Brueggeman]

It sounds to me like you've got old modules hanging around as left-overs.
Did you build LVM 0.5 with the same kernel version? (2.2.2 presumably).  If
so, did you clear out your old modules directory before you rebuilt the
kernel with LVM 0.6?  Are you using modversions when you build the
kernel/modules?

I've also noticed that vgscan will cause the kernel will complain about not
being able to find some block devices that I have not compiled into
modules, or the kernel.  This doesn't sound like your problem though.

I've also had a problem when vgscan ran, because of a bug in the md driver.
I run RedHat 5.2, and the default installation created /dev/md0 through
/dev/md15, but the default md driver is compiled to handle only 4 md
devices (#define MAX_MD_DEV  4), and because of a bug in the md_open()
(uses [minor] without checking minor is less than MAX_MD_DEV), my kernel
would crash when vgscan tried to check /dev/md6 or so.  I just rm'd
/dev/md[4-9] /dev/md1[0-5] to not cause the problem.   I don't think this
is your problem either, but thought you might like to know.

Steve Brueggeman
(All opinions are my own, and in no way reflect those of my employer).
(<--legal-crud)





"Verloove, Olivier" <olivier_verloove@merck.com> on 03/11/99 07:40:42 AM

To:   linux-lvm@msede.com
cc:    (bcc: Steve J Brueggeman)
Subject:  RE: [linux-lvm] Bonehead problem




I have the same problem (Linux 2.2.2 + LVM 0.6).
With or without lvm module loaded, I get:
     vgscan -- invalid i/o protocol version

Is there another module which should be loaded at the same time ?
lsmod gives me:
     lvm  26960     0  (unused)

Thanks.
Olivier Verloove

> ----------
> From:
> ishamael@erroneous.demon.co.uk[SMTP:ishamael@erroneous.demon.co.uk]
> Sent:   Thursday, 11 Mar, 1999 1:25
> To:     linux-lvm@msede.com
> Subject:     Re: [linux-lvm] Bonehead problem
>
> Shawn Leas wrote:
> >When I upgraded to v0.6 from v0.5alpha, I end up having
> >to keep the old v0.5 version of vgscan around.
> >
> >Reason? Well, this is WEIRD... v0.6 vgscan always errs out
> >with 96, or, invalid IO/protocol version. Then, I use v0.5
> >vgscan, and the same happens. Then, I use v0.6 vgscan again,
> >and IT WORKED!
> >
> >What the hell is going on?
>
> Did you build lvm as a module? _insmod lvm_ worked for me, after several
> hours
> of profound confusion.
>
> -michael
> --
> St michael (mainly) Erroneous   http://www.erroneous.demon.co.uk/
> "It's ok to be bitter and twisted,
>  Provided you don't mind leaving a nasty taste in your own mouth."
>
>

RE: [linux-lvm] Bonehead problem

[Steve_J_Brueggeman]

It sounds to me like you've got old modules hanging around as left-overs.
Did you build LVM 0.5 with the same kernel version? (2.2.2 presumably).  If
so, did you clear out your old modules directory before you rebuilt the
kernel with LVM 0.6?  Are you using modversions when you build the
kernel/modules?

I've also noticed that vgscan will cause the kernel will complain about not
being able to find some block devices that I have not compiled into
modules, or the kernel.  This doesn't sound like your problem though.

I've also had a problem when vgscan ran, because of a bug in the md driver.
I run RedHat 5.2, and the default installation created /dev/md0 through
/dev/md15, but the default md driver is compiled to handle only 4 md
devices (#define MAX_MD_DEV  4), and because of a bug in the md_open()
(uses [minor] without checking minor is less than MAX_MD_DEV), my kernel
would crash when vgscan tried to check /dev/md6 or so.  I just rm'd
/dev/md[4-9] /dev/md1[0-5] to not cause the problem.   I don't think this
is your problem either, but thought you might like to know.

Steve Brueggeman
(All opinions are my own, and in no way reflect those of my employer).
(<--legal-crud)





"Verloove, Olivier" <olivier_verloove@merck.com> on 03/11/99 07:40:42 AM

To:   linux-lvm@msede.com
cc:    (bcc: Steve J Brueggeman)
Subject:  RE: [linux-lvm] Bonehead problem




I have the same problem (Linux 2.2.2 + LVM 0.6).
With or without lvm module loaded, I get:
     vgscan -- invalid i/o protocol version

Is there another module which should be loaded at the same time ?
lsmod gives me:
     lvm  26960     0  (unused)

Thanks.
Olivier Verloove

> ----------
> From:
> ishamael@erroneous.demon.co.uk[SMTP:ishamael@erroneous.demon.co.uk]
> Sent:   Thursday, 11 Mar, 1999 1:25
> To:     linux-lvm@msede.com
> Subject:     Re: [linux-lvm] Bonehead problem
>
> Shawn Leas wrote:
> >When I upgraded to v0.6 from v0.5alpha, I end up having
> >to keep the old v0.5 version of vgscan around.
> >
> >Reason? Well, this is WEIRD... v0.6 vgscan always errs out
> >with 96, or, invalid IO/protocol version. Then, I use v0.5
> >vgscan, and the same happens. Then, I use v0.6 vgscan again,
> >and IT WORKED!
> >
> >What the hell is going on?
>
> Did you build lvm as a module? _insmod lvm_ worked for me, after several
> hours
> of profound confusion.
>
> -michael
> --
> St michael (mainly) Erroneous   http://www.erroneous.demon.co.uk/
> "It's ok to be bitter and twisted,
>  Provided you don't mind leaving a nasty taste in your own mouth."
>
>

Re: [linux-lvm] Bonehead problem

[Klaus Strebel]

Hi Olivier,

"Verloove, Olivier" wrote:
> 
> I have the same problem (Linux 2.2.2 + LVM 0.6).
> With or without lvm module loaded, I get:
>         vgscan -- invalid i/o protocol version
> 
> Is there another module which should be loaded at the same time ?
> lsmod gives me:
>         lvm     26960   0  (unused)
Do you have the char-device /dev/lvm (Major 109, Minor 0) ? The LVM-0.5 module
seems to create this device, if it doesn't exist. The LVM-0.6 seems not to do
this, i use devfs and i have to create the device each time before i load the
module.

Cheers
Klaus
-- 
Klaus Strebel
stb@ep-ag.com
EIGNER + PARTNER AG   - The Engineering Warehouse Company -
<http://www.ep-ag.com>
-----------------------------------------------------------------------
In Heaven there will be no DOS, no Intel, and above all, no Bill Gates.

RE: [linux-lvm] Bonehead problem

[Verloove, Olivier]

I have the same problem (Linux 2.2.2 + LVM 0.6).
With or without lvm module loaded, I get:
	vgscan -- invalid i/o protocol version

Is there another module which should be loaded at the same time ?
lsmod gives me:
	lvm	26960	0  (unused)

Thanks.
Olivier Verloove

> ----------
> From:
> ishamael@erroneous.demon.co.uk[SMTP:ishamael@erroneous.demon.co.uk]
> Sent: 	Thursday, 11 Mar, 1999 1:25
> To: 	linux-lvm@msede.com
> Subject: 	Re: [linux-lvm] Bonehead problem
> 
> Shawn Leas wrote:
> >When I upgraded to v0.6 from v0.5alpha, I end up having
> >to keep the old v0.5 version of vgscan around.
> >
> >Reason? Well, this is WEIRD... v0.6 vgscan always errs out
> >with 96, or, invalid IO/protocol version. Then, I use v0.5
> >vgscan, and the same happens. Then, I use v0.6 vgscan again,
> >and IT WORKED!
> >
> >What the hell is going on?
> 
> Did you build lvm as a module? _insmod lvm_ worked for me, after several
> hours 
> of profound confusion.
> 
> -michael
> -- 
> St michael (mainly) Erroneous   http://www.erroneous.demon.co.uk/
> "It's ok to be bitter and twisted,
>  Provided you don't mind leaving a nasty taste in your own mouth."
> 
> 

Re: [linux-lvm] Bonehead problem

[ishamael]

Shawn Leas wrote:
>When I upgraded to v0.6 from v0.5alpha, I end up having
>to keep the old v0.5 version of vgscan around.
>
>Reason? Well, this is WEIRD... v0.6 vgscan always errs out
>with 96, or, invalid IO/protocol version. Then, I use v0.5
>vgscan, and the same happens. Then, I use v0.6 vgscan again,
>and IT WORKED!
>
>What the hell is going on?

Did you build lvm as a module? _insmod lvm_ worked for me, after several hours 
of profound confusion.

-michael
-- 
St michael (mainly) Erroneous   http://www.erroneous.demon.co.uk/
"It's ok to be bitter and twisted,
 Provided you don't mind leaving a nasty taste in your own mouth."

[linux-lvm] Bonehead problem

[Shawn Leas]

When I upgraded to v0.6 from v0.5alpha, I end up having
to keep the old v0.5 version of vgscan around.

Reason? Well, this is WEIRD... v0.6 vgscan always errs out
with 96, or, invalid IO/protocol version. Then, I use v0.5
vgscan, and the same happens. Then, I use v0.6 vgscan again,
and IT WORKED!

What the hell is going on?

-- Shawn
<=========== America Held Hostage ===========>
   Day 2240 for the poor and the middle class. 
   Day 2259 for the rich and the dead.
   682 days remaining in the Raw Deal.
<============================================>