[linux-audio-user] Journaling filesystems [was: Ardour Crash + can't boot]

[Joey Reid]

I am using xfs on my fedora/ccrma box (kernel from atrpms), but I 
haven't got down to any audio work yet - when i do I'll report my 
results.

On Apr 26, 2004, at 3:06 PM, Jan Depner wrote:

> I would be willing to bet that XFS might be even better than reiserfs
> but I have no data on that.  Mark Knecht documented the responses of 
> the
> different filesystems using Benno Senoner's Latency Test program.  I
> have the results on my site at:
>
> http://myweb.cableone.net/eviltwin69/Arcana.html
>
>
> My own (totally unscientific) results are also commented on there.
>
>
> Jan
>
>
> On Mon, 2004-04-26 at 12:40, Maarten de Boer wrote:
>>> If you're going to rebuild use ext3 instead of ext2 for the root
>>> partition.  It's journalled.  Use Reiserfs for the data partition.  
>>> The
>>
>> What is the reason for using two different filesystems here? Is 
>> reiserfs
>> more suitable for (audio) data?
>>
>> Personally, I am using xfs (also journalled) for all workstations I
>> install, after having used it successfully on a heavy duty fileserver
>> for more than 2 years. It never gave me any problems, and I have done
>> some nasty tests.
>>
>> Now, I have never run Ardour - on top of my TODO list for a looong 
>> time
>> :-), so I can't say how Ardour and xfs play together, but I'd be 
>> surprised
>> if there are any problems. Anyway, I would be very much interested to 
>> hear
>> if others are using xfs, and how it behaves under heavy multitrack 
>> audio
>> IO.
>>
>> Ah, and xfs has a special "realtime" mode. From the kernel 
>> configuration:
>>
>>   If you say Y here you will be able to mount and use XFS filesystems
>>   which contain a realtime subvolume. The realtime subvolume is a
>>   separate area of disk space where only file data is stored. The
>>   realtime subvolume is designed to provide very deterministic
>>   data rates suitable for media streaming applications.
>>
>> which sounds very interesting. BUT!:
>>
>>   This feature is unsupported at this time, is not yet fully
>>   functional, and may cause serious problems.
>>
>> Serious problems... Hmm, looking at Aaron's mail, he already got
>> enough serious problems even without experimental xfs features...
>>
>> xfs is in kernel 2.6.x, and has been included in 2.4.25 as well.
>> (before that, sgi provides patches)
>>
>> maarten
>