🚀 go-pugleaf

Hi Alex,

AFAIK, software mirroring your swap space at best serves no purpose and at 
worst is possibly a dangerous thing to attempt.  Probably better is to use 
both swap partitions as is and let the kernel take advantage of parallelism 
by swapping to both disks simultaneously as it sees fit.

Also be aware that using a single pair of physical disks for two or more 
mirror sets doesn't increase your data safety -- if either disk completely 
dies it'll toast every partition on the disk.  You may well have reasons 
for the partitions, but enhanced redundancy isn't one of them.

As an aside, I was reading through the latest Linux RAID code recently...
There's a mirror read optimization which tries to issue read requests to 
the partition most likely to have the data available soonest, but the code 
is fooled if you also have activity from other partitions on the same 
disk(s).  Not a big deal really.

I'm looking forward to striping my installation in 1Q'02 when I get some 
$$$ for new hardware.  Good luck with your system!

- Jeff


alex wrote:

> hi
> 
> i've tried to create a raid 1 on my suse 7.2. there are 2 identical
> harddisks inside. IDE 40GB IBM Drive. i've installed on the first (hda*)
> the os and than create the partitions with yast on hdb*. all hdb
> partitions have the same size and the same filessystem set.
> 
> hda1 = hdb1 = Linux Native
> hda2 = hdb2 = Linux Swap
> hda3 = hdb3 = Linux Native
> 
> now i created a /etc/raidtab
...
> # Disk2 - raid-1 configuration
> raiddev /dev/md1
>         raid-level      1
>         nr-raid-disks   2
>         nr-spare-disks  0
>         chunk-size      4
>         device          /dev/hda2
>         raid-disk       0
>         device          /dev/hdb2
>         raid-disk       1
> 
...
> 1. now i will run mkraid /dev/hda1 and it will crash
> handling MD device /dev/md0
> analyzing super-block
> disk 0: /dev/hda1, 24066kB, raid superblock at 24000kB
> mkraid: aborted, see the syslog and /proc/mdstat for potential clues.
> 
> 2. now i will run mkraid /dev/hda2 - created sucessfull
> 3. with /dev/hda3 the same problem as /dev/hda1


-- 
  Jeff Cowen     \@ ^\
'98 Rans V-Rex    \==-\-%
  Waltham, MA    ()    O

http://www.BlueSNAFU.com/ -- v2.50