[mythtv-users] Storage Group Error

[Michael T. Dean]

On 03/28/2008 07:12 AM, Tim Phipps wrote:
> Gregory Nordeen wrote:
>> I have 4 300g drives on my mythbox. Here is a listing of the drives.
>>  
>> mythtv at mythtv8a ~]$ df -h
>> Filesystem            Size  Used Avail Use% Mounted on
>> /dev/sda2              19G  5.3G   13G  30% /
>> /dev/sdd1             299G  348M  298G   1% /storage
>> /dev/sdc1             299G   38M  299G   1% /storage1
>> /dev/sdb1             299G   38M  299G   1% /storage2
>> /dev/sda5             277G   35M  277G   1% /storage3
>> /dev/sda1              99M   18M   76M  20% /boot
>> tmpfs                 503M     0  503M   0% /dev/shm
>>     
> ...
>   
>> The total for the system is being reported @ 893G vs 1.2T.
>>  
>> I am at a loss.
> The backend can't just assume that different directories are on 
> different disks. It is possible to bind, nfs or otherwise mount one 
> filesystem in more than one place. The backend looks at the sorage space 
> reported for each directory and assumes any that are identical within a 
> certain amount are the same.

Yep.

>  Often the complaint is that slave backends 
> with nfs-mounted share storage areas are counted as seperate and the 
> space estimate is over. In your case your drives are too similar and the 
> estimate is under.

This is very common for brand new identical drives--but only while
they're still unused.  Once they're used, it's very uncommon for them to
show up as the same filesystem, and when they do, it's transient.

>  There are three options I can think of:
>
> 1) Don't worry. As they get used Myth will realize they are different 
> and restarting the backend will fix it.
>
> 2) Use tune2fs to set slightly different reserved block counts so the 
> sizes look different.
>   

Or:

dd if=/dev/zero of=/storage2/DELETE_ME bs=4M count=250

with a cron job to delete it in a week or just delete it manually after
you get a recording on there or something.

> 3) Reformat them into a RAID5 array before you start using them. 
> Seriously, one of these will fail sooner than you expect, probably when 
> you are away on holiday. You would lose 300GB storage but by the time 
> that matters you could probably ebay those drives and replace them with 
> 4x1TB.

Though using a single filesystem has many disadvantages compared to the
multiple filesystems on multiple spindles (drives) approach (when it
comes to fragmentation, performance (due to seeking), power usage (if
you spin down hard drives), etc.).  IMHO, you should /only/ do RAID with
0.21 if you want redundancy.  (And, IMHO, TV doesn't warrant
redundancy--but I realize others value it differently.)

Mike