Re: raid1 / mdadm issues on reboot - /dev/md* not showing up

[Derek Bosch]

forgot to attach my rcS.d
S02hostname.sh
S02mountkernfs.sh
S04mountdevsubfs.sh
S05bootlogd
S05keymap.sh
S06keyboard-setup
S07hdparm
S08hwclockfirst.sh
S10checkroot.sh
S11hwclock.sh
S12mtab.sh
S18ifupdown-clean
S20module-init-tools
S20policycoreutils
S25mdadm-raid
S30checkfs.sh
S30procps
S35mountall.sh
S36mountall-bootclean.sh
S36udev-mtab
S37mountoverflowtmp
S38pppd-dns
S39ifupdown
S40networking
S43portmap
S44nfs-common
S45mountnfs.sh
S46mountnfs-bootclean.sh
S48console-screen.sh
S49console-setup
S50alsa-utils
S55bootmisc.sh
S55urandom
S70nviboot
S70screen-cleanup
S70x11-common
S75policykit
S75sudo
S99stop-bootlogd-single



On Wed, Aug 26, 2009 at 12:42 PM, Derek Bosch  wrote:

> it appears that if I let the system continue booting, the remaining
> /dev/md*s do get populated, which makes me suspicious of my /etc/rc*.d/
> ordering...
>
> On Wed, Aug 26, 2009 at 12:03 PM, martin f krafft wrote:
>
>> also sprach Derek Bosch  [2009.08.26.2020 +0200]:
>> > md3 : active (auto-read-only) raid1 sda4[0] sdb4[1]
>> > 280631360 blocks [2/2] [UU]
>> >
>> > this device DOESN'T appear in /dev/md3
>> >
>> > however:
>> > md2 : active raid1 sda3[0] sdb3[1]
>> > 9767424 blocks [2/2] [UU]
>> >
>> > isn't auto-read-only, and does appear as /dev/md2...
>> >
>> > I'd like to reset the "auto-read-only" on /dev/md3, but /dev/md3 doesn't
>> > exist.  Sometimes I've seen it show up as /dev/.tmp.md3,
>>
>> File a bug, please.
>>
>> I doubt this has to do with auto-read-only, which is just a symptom
>> because the filesystem probably doesn't get mounted, hence the array
>> is not written and thus stays auto-read-only. The real issue is why
>> the node doesn't get renamed like it should.
>>
>> --
>>  .''`.   martin f. krafft   Related projects:
>> : :'  :  proud Debian developer   http://debiansystem.info
>> `. `'`   
>> http://people.debian.org/~madduck
>> http://vcs-pkg.org
>>  `-  Debian - when you have better things to do than fixing systems
>>
>> all software projects are done by iterative prototyping.
>> some companies call their prototypes "releases", that's all.
>>
>> -BEGIN PGP SIGNATURE-
>> Version: GnuPG v1.4.9 (GNU/Linux)
>>
>> iEYEAREDAAYFAkqVhwgACgkQIgvIgzMMSnXD8gCfcxX2nCpOmjSJ8SftUZXPAZwV
>> E7wAoIfIHwCFShsF1hWb0958Naof0eV9
>> =HGNp
>> -END PGP SIGNATURE-
>>
>>
>

Re: raid1 / mdadm issues on reboot - /dev/md* not showing up

[Derek Bosch]

it appears that if I let the system continue booting, the remaining
/dev/md*s do get populated, which makes me suspicious of my /etc/rc*.d/
ordering...

On Wed, Aug 26, 2009 at 12:03 PM, martin f krafft wrote:

> also sprach Derek Bosch  [2009.08.26.2020 +0200]:
> > md3 : active (auto-read-only) raid1 sda4[0] sdb4[1]
> > 280631360 blocks [2/2] [UU]
> >
> > this device DOESN'T appear in /dev/md3
> >
> > however:
> > md2 : active raid1 sda3[0] sdb3[1]
> > 9767424 blocks [2/2] [UU]
> >
> > isn't auto-read-only, and does appear as /dev/md2...
> >
> > I'd like to reset the "auto-read-only" on /dev/md3, but /dev/md3 doesn't
> > exist.  Sometimes I've seen it show up as /dev/.tmp.md3,
>
> File a bug, please.
>
> I doubt this has to do with auto-read-only, which is just a symptom
> because the filesystem probably doesn't get mounted, hence the array
> is not written and thus stays auto-read-only. The real issue is why
> the node doesn't get renamed like it should.
>
> --
>  .''`.   martin f. krafft   Related projects:
> : :'  :  proud Debian developer   http://debiansystem.info
> `. `'`   
> http://people.debian.org/~madduck
> http://vcs-pkg.org
>  `-  Debian - when you have better things to do than fixing systems
>
> all software projects are done by iterative prototyping.
> some companies call their prototypes "releases", that's all.
>
> -BEGIN PGP SIGNATURE-
> Version: GnuPG v1.4.9 (GNU/Linux)
>
> iEYEAREDAAYFAkqVhwgACgkQIgvIgzMMSnXD8gCfcxX2nCpOmjSJ8SftUZXPAZwV
> E7wAoIfIHwCFShsF1hWb0958Naof0eV9
> =HGNp
> -END PGP SIGNATURE-
>
>

Re: raid1 / mdadm issues on reboot - /dev/md* not showing up

[martin f krafft]

also sprach Derek Bosch  [2009.08.26.2020 +0200]:
> md3 : active (auto-read-only) raid1 sda4[0] sdb4[1]
> 280631360 blocks [2/2] [UU]
> 
> this device DOESN'T appear in /dev/md3
> 
> however:
> md2 : active raid1 sda3[0] sdb3[1]
> 9767424 blocks [2/2] [UU]
> 
> isn't auto-read-only, and does appear as /dev/md2...
> 
> I'd like to reset the "auto-read-only" on /dev/md3, but /dev/md3 doesn't
> exist.  Sometimes I've seen it show up as /dev/.tmp.md3,

File a bug, please.

I doubt this has to do with auto-read-only, which is just a symptom
because the filesystem probably doesn't get mounted, hence the array
is not written and thus stays auto-read-only. The real issue is why
the node doesn't get renamed like it should.

-- 
 .''`.   martin f. krafft   Related projects:
: :'  :  proud Debian developer   http://debiansystem.info
`. `'`   http://people.debian.org/~madduckhttp://vcs-pkg.org
  `-  Debian - when you have better things to do than fixing systems
 
all software projects are done by iterative prototyping.
some companies call their prototypes "releases", that's all.


digital_signature_gpg.asc
Description: Digital signature (see http://martin-krafft.net/gpg/)

raid1 / mdadm issues on reboot - /dev/md* not showing up

[Derek Bosch]

I've had a sid/unstable system with four raid1 / mdadm partitions running
for about 4 years now...
Recently, upon boot-up, three of those four partitions come up as
"auto-read-only" in /proc/mdstat
for example:
md3 : active (auto-read-only) raid1 sda4[0] sdb4[1]
280631360 blocks [2/2] [UU]

this device DOESN'T appear in /dev/md3

however:
md2 : active raid1 sda3[0] sdb3[1]
9767424 blocks [2/2] [UU]

isn't auto-read-only, and does appear as /dev/md2...

I'd like to reset the "auto-read-only" on /dev/md3, but /dev/md3 doesn't
exist.  Sometimes I've seen it show up as /dev/.tmp.md3,
but it isn't consistent...

where does it go?

-Derek