[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: automatically get the server up



Umum Wijoyo wrote:

>
>Hi!
>I'd like to ask why my Debian server, on a power-down or black-out, 
>always requests for root to do an fsck etc. 
>Is there any way I can automatically get the server up and running again? 
>Is this a good solution? Or is it better off having root 
>to do a manual fsck etc first?
>Thanks!


That's one of the disadvantages of ext2fs -- the long fsck especially 
after power failure. This is done to preserve the integrity of the 
filesystem. It's not a good idea to tune2fs -c0 -i0 /dev/hdX the system 
to avoid the long fsck's for you might bump into trouble later, 
especially when power failures would force the fsck's.

If you're really want to avoid the long fsck's, try converting your 
filesystem to journalling filesystems, like ext3fs (which is quite easy, 
as it's just ext2 with a journal file - try installing the e2fsprogs-
1.22 or up packages from sid then run tune2fs -j /dev/hdX) or the likes 
of XFS or reiserfs. By experience, I prefer reiserfs, as it integrates 
seamlessly with ext3fs (which is the root filesystem on my Debian 
server, as opposed to XFS). However, reiserfs has some problems with NFS 
(as it doesn't use inodes... ) and I've yet to succeed installing lilo 
with a reiserfs partition as root. 

XFS is good if you're dealing with very big files in the filesystem (it was developed for big files such as computer graphics/ movies) Reiserfs
partitions can theoretically be as big as your hard disk and has a very fast delete rate. Ext3 proposes full journalling of data instead of just mere meta-data; however it inherits the shortcomings of ext2, especially when you run out of inodes. Go figure what you want and really need.

Paolo Falcone

__________________________________
www.edsamail.com



Reply to: