Hi Group ,
I need your suggestions in configuring RAID - 5 for file servers ,
firstly i want to know which tools(open source s/w) should be used
for managing RAID with less complexity,more flexiblity and support for
resizing data volume(without losing of data integrity), Currently i
am using 'mdm' and 'raidtools' but these don't solve my purpose. What
do you think about 'EVMS's' functionalities and flexiblities ? which
filesystem is more stable and easy to recover in case of disk failure
(ext3,JFS,ReiserFS,Open GFS)?

how can i configure RAID-1 in such a way to continue booting the system
in degraded mode even after failing of a disk . Is storing the /boot
partition in RAID-1 is harmful or it should be located on another
partition.

well i have many queries related to RAID ,
There is anyone who can solve my problems.

Thanks in advance

Regards
Ritesh Agrawal

I use RAID-1 / ext3 for my /boot partition. I normally would create the partition at the beginning of the disks.

I do not mount the /boot partition unless it's needed for a kernel upgrade.


/etc/fstab


LABEL=/boot /boot ext3 noauto,owner 1 2


lsraid -a /dev/md1
[dev 9, 1] /dev/md1 64012470.1FDA2CA2.2C060F8F.3EE232AD online
[dev 3, 1] /dev/hda1 64012470.1FDA2CA2.2C060F8F.3EE232AD good
[dev 3, 65] /dev/hdb1 64012470.1FDA2CA2.2C060F8F.3EE232AD good


# md device [dev 9, 1] /dev/md1 queried online
raiddev /dev/md1
raid-level 1
nr-raid-disks 2
nr-spare-disks 0
persistent-superblock 1
chunk-size 64

device /dev/hda1
raid-disk 0
device /dev/hdb1
raid-disk 1



I have not had any problems other than a bad hard drive. The system can still boot if you can boot from the other disk, you installed the boot loader to both disks and the / filesystem is also RAID. Or using a boot floppy. I usually have the system email me in case of a problem.



cat raidchk.cron
#!/bin/sh
if [ "`lsraid -A -f -a /dev/md0 -a /dev/md1 | grep failed`" ]
then
lsraid -a /dev/md0 -a /dev/md1 | mail -s "Raid array problem" root
fi

Keep the file storage hard drives away or seperate from your OS hard drives. This way if the raid array goes bad you can easily boot into LINUX and find out why the raid array is bad.

You can use EVMS if you want to add more storage later. With this feature in the kernel, something may go bad in the future.

ext3 is an journalizing file system. It is easy to fix and it can go back and forth from ext3 to ext2.

reiserfs is an journalizing file system. It has different algorithms when storing its journal for better performance or relaiblity. Great performance for small files and it can store several times more than ext3 in a directory with out losing performance.

jfs is an journalizing file system from IBM. Great high performance but becomes sloppy when saving files. Needs defrag every few weeks or months although does not come with a defrag utility. User have to dump the partition to an image and then back on to the drive. People said its great for working with OS files.

XFS is an journalizing file system from SGI with interesting features up its sleeves. The journal can be located on another drive. If one hard drive fails. XFS can recreate the data or journal from the other drive. It has very, very good performance with big files and serveral thousand files in a directory. The disadvantage of using XFS is you can not use lilo. If you become accustom to lilo I'm sorry you have to move to grub. This is because XFS writes to the superblock where lilo puts some its data.

Using software raid is ok on a workstation, but its not on a server. Its best to go with hardware raid. If you want to use the hardware that you got, you can use devices from Accusys. They make it easy for the user when setting up hardware RAID 0, 1, 5 for IDE or SCSI. With Accuys you can transport it to another system with out damaging the array.

RAID doesn't help much so you have to backup at regular intervals and send the backups to a data service that is out of town.