[NTLUG:Discuss] strange "No space" issue on /dev/md0

[terry]

TJ Davis wrote:

>
> Yes, it is saying that I had to unmount it which would not be worth my 
> effort because it is an active server.  I was just wondering if I 
> could do a passive check on it without unmounting it that way if it is 
> bad I can unmount it and clean it up.  The -N switch did not do the 
> trick by the way.
>
I've yet to run fsck on a mounted partition, so I cannot answer your 
question on that.  
All I know for sure is that it's SOP to unmount a particular filesystem 
before running fsck, and there are probably a couple of reasons for that 
recommendation.  

* Probably the main reason is "dirty buffers", or the need to "sync the 
disk".  Syncing the disk is normally done during the process of 
un-mounting a filesystem.  Syncing the disk and/or flushing the buffers 
can be accomplished [while filesystem is mounted] by running, sync. 
 (Running sync not just once but three times  is recommended to ensure 
that all the buffers are properly flushed to disk.)

If you're upgrading from RH 7.1 to RH7.3, that's a good thing, 
especially since  you'll be able to run ext3.  fsck runs much faster on 
ext3, * ("typically less than 10 percent the time it takes to check a 
nonjournaling filesystem.")

*Info source: "Using Linux Sixth Edition", by David Bandel & Robert Napier