[Ocfs2-users] Lots of questions & lustre

[Sunil Mushran]

Jeffrey W. Baker wrote:

>On a 4TB volume with 4KB blocks, the maximum number of inodes will be 1
>billion?
>
>  
>
Close. The 4TB will also include space to allocate the system files
like journals for each node, bitmap global and local, accounting for the 
inode
themselves. So, you'll need to remove space for all that before doing 
the arithmetic.

>I ask this question because ext2/ext3 have a limit of 32,000
>subdirectories.
>
>  
>
Yes, we have the same limit.... considering the directory handling
in ocfs2 is based upon ext2/3's. :)

 #define OCFS2_LINK_MAX          32000

      if (S_ISDIR(mode) && (dir->i_nlink >= OCFS2_LINK_MAX)) {
                mlog(ML_ERROR, "inode %"MLFu64" has i_nlink of %u\n",
                     OCFS2_I(dir)->ip_blkno, dir->i_nlink);
                status = -EMLINK;
                goto leave;
        }

However, we could safely bump the number to 65000 as i_nlink is u16
in ocfs2.

>>>Can you safely deliver mail on OCFS2?
>>>
>>>      
>>>
>>Care to expand.
>>    
>>
>
>Mail delivery is a kind of acid test of filesystem usefulness.  You
>cannot safely deliver mail without an atomic link(2) or rename(2).  For
>example, you cannot safely deliver mail on NFS.
>
>  
>
That we do. However, we currently don't support posix locking across the 
cluster.
Try it out and let us know what issues you encounter.