Re: Linux ready for the home desktop?

["Ben Eisenbraun" <user=82@nabble.blu.org>]

 On Thu, Mar 27, 2008 at 01:28:45PM -0400, Dan Ritter wrote: 
> On Thu, Mar 27, 2008 at 01:03:56PM -0400, Ben Eisenbraun wrote: 
> > Why is this the default in the first place?  Is it just a hold-over from 
> > the days when ext2 really sucked?  Windows, Mac OS and the BSDs don't do 
> > file system checks unless the file system is marked unclean. 
> 
> It's a feature from the time when computers, software and electricity 
> were not 100% reliable and consistent. 

Well, that applies even today, but maybe that's what you're getting at. 
  
> > Maybe it was a crutch supporting the original ext2 philosophy of 
> > speed-over-sanity. 
> 
> La, an inflammatory remark without supporting evidence. 

I didn't realize it was inflammatory.  ext2 uses async mounts by default, 
so under heavy i/o load, it's possible for the buffer cache to have 
unordered, in-progress file system metadata updates which would not be 
recoverable after a power loss/crash/etc.  If there is extensive enough 
corruption, then you are presented with the opportunity to test your backups. 

I kinda thought this was a well-known trade off made by the ext2 folks, 
hence my speed-vs-sanity comment. 

http://www.usenix.org/publications/library/proceedings/usenix2000/general/seltzer.html

The linked paper has some info on the trade-offs between sync and async 
mounts, and the impetus for journaling or using a method like soft updates 
to maintain the integrity of the file system metadata. 

-b 

-- 
ignorance is always ready to admire itself.  procure yourself critical 
friends.                                             <nicolas boileau> 

-- 
This message has been scanned for viruses and 
dangerous content by MailScanner, and is 
believed to be clean. 

_______________________________________________ 
Discuss mailing list 
[hidden email] 
http://lists.blu.org/mailman/listinfo/discuss