SLES 9 SP3 and mdadm 2.6.1 (via rpm)

SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Thorsten Wolf]

Hello everyone.

I've got a SLES9 SP3 running and I've been quite happy with it so far.

Recently, I've created a 4 disk spanning RAID-5 on our company server. Runs quite nice and we're happy with that too. I created that RAID using the SLES mdadm (1.4 I believe) package.

After discovering that there is a much newer mdadm out here (2.6.1), I decided to upgrade. It went just fine. Raid still running at 120 MB/sec.

After adding a disk to the raid, which went fine as well...... BUT:

The added disk /dev/sda1 shows up in /proc/mdstat, but does not have the "spare (s)" flag.

Plus... the --grow doesn't work...

I get the: mdadm: /dev/md0: Cannot get array details from sysfs error which has been discussed before. Can it be that this is caused by the 2.6.5-7.2xx Kernel? Any ideas?

regards,

Thorsten
-- 
Contact me on ICQ: 7656468
skype://sysfried

GMX FreeMail: 1 GB Postfach, 5 E-Mail-Adressen, 10 Free SMS.
Alle Infos und kostenlose Anmeldung: http://www.gmx.net/de/go/freemail

Re: SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Neil Brown]

On Tuesday June 12, T_Wolf@gmx.net wrote:
> Hello everyone.
> 
> I've got a SLES9 SP3 running and I've been quite happy with it so far.
> 
> Recently, I've created a 4 disk spanning RAID-5 on our company
> server. Runs quite nice and we're happy with that too. I created
> that RAID using the SLES mdadm (1.4 I believe) package. 
> 
> After discovering that there is a much newer mdadm out here (2.6.1),
> I decided to upgrade. It went just fine. Raid still running at 120
> MB/sec. 
> 
> After adding a disk to the raid, which went fine as well...... BUT:
> 
> The added disk /dev/sda1 shows up in /proc/mdstat, but does not have
> the "spare (s)" flag. 
> 
> Plus... the --grow doesn't work...
> 
> I get the: mdadm: /dev/md0: Cannot get array details from sysfs
> error which has been discussed before. Can it be that this is caused
> by the 2.6.5-7.2xx Kernel? Any ideas? 

Yes.  All of your issues are caused by using a 2.6.5 based kernel.
However even upgrading to SLES10 would not get you raid5-grow.  That
came a little later.  You would need to compile a mainline kernel or
wait for SLES11.

NeilBrown

Re: SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Bill Davidsen]

Neil Brown wrote:
> On Tuesday June 12, T_Wolf@gmx.net wrote:
>   
>> Hello everyone.
>>
>> I've got a SLES9 SP3 running and I've been quite happy with it so far.
>>
>> Recently, I've created a 4 disk spanning RAID-5 on our company
>> server. Runs quite nice and we're happy with that too. I created
>> that RAID using the SLES mdadm (1.4 I believe) package. 
>>
>> After discovering that there is a much newer mdadm out here (2.6.1),
>> I decided to upgrade. It went just fine. Raid still running at 120
>> MB/sec. 
>>
>> After adding a disk to the raid, which went fine as well...... BUT:
>>
>> The added disk /dev/sda1 shows up in /proc/mdstat, but does not have
>> the "spare (s)" flag. 
>>
>> Plus... the --grow doesn't work...
>>
>> I get the: mdadm: /dev/md0: Cannot get array details from sysfs
>> error which has been discussed before. Can it be that this is caused
>> by the 2.6.5-7.2xx Kernel? Any ideas? 
>>     
>
> Yes.  All of your issues are caused by using a 2.6.5 based kernel.
> However even upgrading to SLES10 would not get you raid5-grow.  That
> came a little later.  You would need to compile a mainline kernel or
> wait for SLES11.

I have to think that if features require a later kernel version that a 
warning message would be appropriate. I'm always leary about trying a 
new mdadm version with a vendor kernel unless it's a minor bugfix problem.

-- 
bill davidsen <davidsen@tmr.com>
  CTO TMR Associates, Inc
  Doing interesting things with small computers since 1979

Re: SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Thorsten Wolf]

Dear Neil,
hi everyone else.

I've been reading a lot about the mdadm tools lately, and I believe that it is possible, but I haven't found the right documentation yet.

Is it possible to re-mount the RAID I have from another Linux installation? Do I need more than the 'mdadm --detail --scan > /etc/mdadm.conf' info saved to another location (so I can re-import it)?

Regards,

Thorsten

> On Tuesday June 12, T_Wolf@gmx.net wrote:
> > Hello everyone.
> > 
> > I've got a SLES9 SP3 running and I've been quite happy with it so far.
> > 
> > Recently, I've created a 4 disk spanning RAID-5 on our company
> > server. Runs quite nice and we're happy with that too. I created
> > that RAID using the SLES mdadm (1.4 I believe) package. 
> > 
> > After discovering that there is a much newer mdadm out here (2.6.1),
> > I decided to upgrade. It went just fine. Raid still running at 120
> > MB/sec. 
> > 
> > After adding a disk to the raid, which went fine as well...... BUT:
> > 
> > The added disk /dev/sda1 shows up in /proc/mdstat, but does not have
> > the "spare (s)" flag. 
> > 
> > Plus... the --grow doesn't work...
> > 
> > I get the: mdadm: /dev/md0: Cannot get array details from sysfs
> > error which has been discussed before. Can it be that this is caused
> > by the 2.6.5-7.2xx Kernel? Any ideas? 
> 
> Yes.  All of your issues are caused by using a 2.6.5 based kernel.
> However even upgrading to SLES10 would not get you raid5-grow.  That
> came a little later.  You would need to compile a mainline kernel or
> wait for SLES11.
> 
> NeilBrown

-- 
Contact me on ICQ: 7656468
skype://sysfried

Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! 
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Justin Piszcz]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 3004 bytes --]

You don't even need that, just do this:

1. echo 'DEVICE /dev/hd*[0-9] /dev/sd*[0-9]' > mdadm.conf
2. mdadm  --examine  --scan --config=mdadm.conf

This will search all partitions and give the relevant SW raid information:

ARRAY /dev/md/4 level=raid5 metadata=1 num-devices=5 
UUID=7f453e1889:3e4dd96e:8103724c:724f49 name=4
ARRAY /dev/md/3 level=raid5 metadata=1 num-devices=4 
UUID=b3d7134904:52828f3d:0f0245a2:e8226d name=3

Then edit these lines so they look like this:

ARRAY /dev/md3 level=raid5 metadata=1 num-devices=4 
UUID=b3d7134904:52828f3d:0f0245a2:e8226d name=3
ARRAY /dev/md4 level=raid5 metadata=1 num-devices=5 
UUID=7f453e1889:3e4dd96e:8103724c:724f49 name=4

Then run:

1. mdadm -As /dev/md3
2. mdadm -As /dev/md4

p34:/etc/mdadm# mdadm -As /dev/md3
mdadm: /dev/md3 has been started with 4 drives.
p34:/etc/mdadm# mdadm -As /dev/md4
mdadm: /dev/md4 has been started with 5 drives.

Done.


On Wed, 13 Jun 2007, Thorsten Wolf wrote:

> Dear Neil,
> hi everyone else.
>
> I've been reading a lot about the mdadm tools lately, and I believe that it is possible, but I haven't found the right documentation yet.
>
> Is it possible to re-mount the RAID I have from another Linux installation? Do I need more than the 'mdadm --detail --scan > /etc/mdadm.conf' info saved to another location (so I can re-import it)?
>
> Regards,
>
> Thorsten
>
>> On Tuesday June 12, T_Wolf@gmx.net wrote:
>>> Hello everyone.
>>>
>>> I've got a SLES9 SP3 running and I've been quite happy with it so far.
>>>
>>> Recently, I've created a 4 disk spanning RAID-5 on our company
>>> server. Runs quite nice and we're happy with that too. I created
>>> that RAID using the SLES mdadm (1.4 I believe) package.
>>>
>>> After discovering that there is a much newer mdadm out here (2.6.1),
>>> I decided to upgrade. It went just fine. Raid still running at 120
>>> MB/sec.
>>>
>>> After adding a disk to the raid, which went fine as well...... BUT:
>>>
>>> The added disk /dev/sda1 shows up in /proc/mdstat, but does not have
>>> the "spare (s)" flag.
>>>
>>> Plus... the --grow doesn't work...
>>>
>>> I get the: mdadm: /dev/md0: Cannot get array details from sysfs
>>> error which has been discussed before. Can it be that this is caused
>>> by the 2.6.5-7.2xx Kernel? Any ideas?
>>
>> Yes.  All of your issues are caused by using a 2.6.5 based kernel.
>> However even upgrading to SLES10 would not get you raid5-grow.  That
>> came a little later.  You would need to compile a mainline kernel or
>> wait for SLES11.
>>
>> NeilBrown
>
> -- 
> Contact me on ICQ: 7656468
> skype://sysfried
>
> Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
> Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
> -
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Luca Berra]

On Wed, Jun 13, 2007 at 07:50:06AM -0400, Justin Piszcz wrote:
>You don't even need that, just do this:
>
>1. echo 'DEVICE /dev/hd*[0-9] /dev/sd*[0-9]' > mdadm.conf
do _NOT_ do the above, _never_.
recent mdadm do not need the "DEVICE" line
for older one use "DEVICE partitions".

-- 
Luca Berra -- bluca@comedia.it
        Communication Media & Services S.r.l.
 /"\
 \ /     ASCII RIBBON CAMPAIGN
  X        AGAINST HTML MAIL
 / \

Re: SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Justin Piszcz]

On Thu, 14 Jun 2007, Luca Berra wrote:

> On Wed, Jun 13, 2007 at 07:50:06AM -0400, Justin Piszcz wrote:
>> You don't even need that, just do this:
>> 
>> 1. echo 'DEVICE /dev/hd*[0-9] /dev/sd*[0-9]' > mdadm.conf
> do _NOT_ do the above, _never_.
> recent mdadm do not need the "DEVICE" line
> for older one use "DEVICE partitions".
>
> -- 
> Luca Berra -- bluca@comedia.it
>       Communication Media & Services S.r.l.
> /"\
> \ /     ASCII RIBBON CAMPAIGN
> X        AGAINST HTML MAIL
> / \
> -
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Why never do that? It has always worked for me.

Justin.

need help in recovering a RAID

[Thorsten Wolf]

Hello again, everyone.

I've been really busy playing and working with mdadm tools lately. And I've come across a problem which I haven't been able to solve myself so far.

My system looks like this:
IDE:
/dev/hda1   -->   /      100 GB
/dev/hda2   -->   SWAP     1 GB
/dev/hda3   -->   /home  395 GB

S-ATA onboard:
/dev/hde1   -->   Disk 0 on /dev/md0
/dev/hdg1   -->   Disk 1 on /dev/md0
/dev/hdi1   -->   Disk 2 on /dev/md0
/dev/hdk1   -->   Disk 3 on /dev/md0

/dev/md0    --> via LVG mounted as /raid

A loose cable caused hdi and hdk to go offline. The server itself was shut down 1 minute later.

But I can't access the raid anymore because it marked the disks as failed. There was no write access to the system when the power went down on the two disks.

How can I mark them as online (I'm pretty sure the data is still ok) ???

I really want to get the raid working again as I think this is a good scenario that I'd like to be able to handle in the future.

Regards,

Thorsten
-- 
Contact me on ICQ: 7656468
skype://sysfried

Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! 
Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Luca Berra]

On Thu, Jun 14, 2007 at 04:22:03AM -0400, Justin Piszcz wrote:
>
>
>On Thu, 14 Jun 2007, Luca Berra wrote:
>
>>On Wed, Jun 13, 2007 at 07:50:06AM -0400, Justin Piszcz wrote:
>>>You don't even need that, just do this:
>>>
>>>1. echo 'DEVICE /dev/hd*[0-9] /dev/sd*[0-9]' > mdadm.conf
>>do _NOT_ do the above, _never_.
>>recent mdadm do not need the "DEVICE" line
>>for older one use "DEVICE partitions".
>>
>
>Why never do that? It has always worked for me.
>
the keyword here is "for me"
but /dev/hd*[0-9] /dev/sd*[0-9] does not mean "all disks" for everybody.

L.

-- 
Luca Berra -- bluca@comedia.it
        Communication Media & Services S.r.l.
 /"\
 \ /     ASCII RIBBON CAMPAIGN
  X        AGAINST HTML MAIL
 / \

Re: SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Thorsten Wolf]

Hello again.

I've upgraded my SLES 9 SP3 system to SLES 10 (no SP1). The raid I had running on my devices:
/dev/hdg1   /dev/hdi1   /dev/hdk1   /dev/hde1
doesn't work because SLES 10 does detect them as:
/dev/sda1   /dev/sdb1   /dev/sdc1   /dev/sdd1


I guess it's going to be simple, but can anyone give me the right syntax to fix this?

Regards,

Thorsten Wolf

-- 
Contact me on ICQ: 7656468
skype://sysfried

Psssst! Schon vom neuen GMX MultiMessenger gehört?
Der kanns mit allen: http://www.gmx.net/de/go/multimessenger
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: SLES 9 SP3 and mdadm 2.6.1 (via rpm)

[Justin Piszcz]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 893 bytes --]



On Mon, 25 Jun 2007, Thorsten Wolf wrote:

> Hello again.
>
> I've upgraded my SLES 9 SP3 system to SLES 10 (no SP1). The raid I had running on my devices:
> /dev/hdg1   /dev/hdi1   /dev/hdk1   /dev/hde1
> doesn't work because SLES 10 does detect them as:
> /dev/sda1   /dev/sdb1   /dev/sdc1   /dev/sdd1
>
>
> I guess it's going to be simple, but can anyone give me the right syntax to fix this?
>
> Regards,
>
> Thorsten Wolf
>
> -- 
> Contact me on ICQ: 7656468
> skype://sysfried
>
> Psssst! Schon vom neuen GMX MultiMessenger gehört?
> Der kanns mit allen: http://www.gmx.net/de/go/multimessenger
> -
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

mdadm --examine --scan and use that in your mdadm.conf ?