[Ocfs2-users] mounted.ocfs2 -f return Unknown: Bad magic number in inode

Thu Jan 15 01:40:35 PST 2009

Please forget my last question about mixed name of device mapper (dm-0 and
dm-4)
dm-0 is linked to sda and sdf
dm-4 is linked to sda1 and sdf1

So just the "bad magic" message disturbs me.

Mohamed.

             Mohamed                                                       
             BOUTKHIL--CAPGEMI                                             
             NI--/MAIL                                                   A 
                                       Sunil Mushran                       
             15/01/2009 10:19          <sunil.mushran at oracle.com>          
                                                                        cc 

                                                                     Objet 
                                       Re: [Ocfs2-users] mounted.ocfs2 -f  
                                       return Unknown: Bad magic number in 
                                       inode(Document link: Mohamed        
                                       BOUTKHIL--CAPGEMINI--)              

Thanks for your quick response.
I used bvi as you suggested and it works fine.
here is what I have now :
 mounted.ocfs2 -f
Device                FS     Nodes
/dev/sda              ocfs2  Unknown: Bad magic number in inode
/dev/sda1             ocfs2  pocrhel2, pocrhel1
/dev/sdf              ocfs2  Unknown: Bad magic number in inode
/dev/sdf1             ocfs2  pocrhel2, pocrhel1
/dev/dm-0             ocfs2  Unknown: Bad magic number in inode
/dev/dm-4             ocfs2  pocrhel2, pocrhel1

Is it normal to have this message : "Unknown: Bad magic number in inode" to
the device where an OCFS2 partition is built ?

And eventually, there is a switch between multipath. Normally dm-0 path is
linked to device sda and sdf (as multipath -ll command says). dm-4 is an
old one no more used
(in fact linked with sdb and sdg that I modified just before with bvi)

Mohamed.

             Sunil Mushran                                                 
             <sunil.mushran at or                                             
             acle.com>                                                   A 
                                       "BOUTKHIL--CAPGEMINI--, Mohamed"    
             13/01/2009 21:03          <mohamed.boutkhil-capgemini at ca-leas 
                                       ing.fr>                             
                                                                        cc 
                                       ocfs2-users at oss.oracle.com          
                                                                     Objet 
                                       Re: [Ocfs2-users] mounted.ocfs2 -f  
                                       return Unknown: Bad magic number in 
                                       inode                               

Appears you formatted the devices before you partitioned them.
As the ocfs2 superblock is in the third block, it is not overwritten
during partition. This version of mounted.ocfs2 does not handle
this situation.

Easiest fix is hand editing the signature using a binary editor.

$ bvi -s 10240 /dev/sda

Change the ocfs2 signature (OCFSV2) to something else. Changing
one character is enough.

BOUTKHIL--CAPGEMINI--, Mohamed wrote:
> Hello,
>
> I have installed ocfs2 without problem and use it for a RAC10gR2.
> Only Clusterware files are ocfs2 type.
> multipath is also used.
> When I issue : mounted.ocfs2 -f
> I have a strange result:
>
> Device                FS     Nodes
> /dev/sda              ocfs2  Unknown: Bad magic number in inode
> /dev/sda1             ocfs2  pocrhel2, pocrhel1
> /dev/sdb              ocfs2  Not mounted
> /dev/sdf              ocfs2  Unknown: Bad magic number in inode
> /dev/sdf1             ocfs2  pocrhel2, pocrhel1
> /dev/sdg              ocfs2  Not mounted
> /dev/dm-0             ocfs2  Unknown: Bad magic number in inode
> /dev/dm-4             ocfs2  pocrhel2, pocrhel1
>
> Devices: sda and sdf  are different paths to the same LUN. dm-0 is the
> multipath as stated by the following command:
>  multipath -ll
> ASM_DISK_4 (360060e80056583000000658300004064) dm-3 HITACHI,OPEN-V
> [size=9.0G][features=0][hwhandler=0]
> \_ round-robin 0 [prio=2][active]
>  \_ 0:0:1:4 sde 8:64  [active][ready]
>  \_ 1:0:1:4 sdj 8:144 [active][ready]
> ASM_DISK_3 (360060e80056583000000658300004063) dm-2 HITACHI,OPEN-V
> [size=9.0G][features=0][hwhandler=0]
> \_ round-robin 0 [prio=2][active]
>  \_ 0:0:1:3 sdd 8:48  [active][ready]
>  \_ 1:0:1:3 sdi 8:128 [active][ready]
> ASM_DISK_2 (360060e80056583000000658300004062) dm-1 HITACHI,OPEN-V
> [size=9.0G][features=0][hwhandler=0]
> \_ round-robin 0 [prio=2][active]
>  \_ 0:0:1:2 sdc 8:32  [active][ready]
>  \_ 1:0:1:2 sdh 8:112 [active][ready]
> CRS_DISK (360060e80056583000000658300004060) dm-0 HITACHI,OPEN-V
> [size=9.0G][features=0][hwhandler=0]
> \_ round-robin 0 [prio=2][active]
>  \_ 0:0:1:0 sda 8:0   [active][ready]
>  \_ 1:0:1:0 sdf 8:80  [active][ready]
>
> I have some questions:
> 1- The message "Unknown: Bad magic number in inode", what does it mean ?
> Can I discard it ?
> 2- sdb and sdg devices were an ocfs2 volume but not anymore. Why it is
> listed here ?
> 3- It seems there is a mismatch between dm-0 and dm-4. dm-0 must be the
> right one. dm-4 is an old multipath removed.. Is this info is reliable ?
>
> Does anyone can help me on this weird behavior?
>
> Thanks in advance,
>
>
> Mohamed.
>
>
> _______________________________________________
> Ocfs2-users mailing list
> Ocfs2-users at oss.oracle.com
> http://oss.oracle.com/mailman/listinfo/ocfs2-users
>