* [linux-lvm] lvm on software raid 5
@ 2004-02-29 23:03 Clemens Schwaighofer
0 siblings, 0 replies; 3+ messages in thread
From: Clemens Schwaighofer @ 2004-02-29 23:03 UTC (permalink / raw)
To: linux-lvm
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
hi,
I have a software raid 5 in linud 2.6.3 and I created a lvm on. Actually
it works, but there some strange things. Perhaps someone can point me
out why.
Basic:
lvm vesion:
ii lvm2 2.00.08-4 The Linux Logical Volume Manager
ii lvm-common 1.5.12 The Logical Volume Manager for
Linux (common files
raid version:
ii raidtools2 1.00.3-5 Utilities to support 'new-style'
RAID disks
md admin version
ii mdadm 1.4.0-3 Manage MD devices aka Linux
Software Raid
my raidtab:
# 4 x 160GB Raid 5, no spare
raiddev /dev/md0
raid-level 5
nr-raid-disks 4
nr-spare-disks 0
persistent-superblock 1
chunk-size 64
parity-algorithm left-symmetric
# Spare disks for hot reconstruction
# no the raid devices
# Alpha
device /dev/sda1
raid-disk 0
# Beta
device /dev/sdb1
raid-disk 1
# Gamma
device /dev/sdc1
raid-disk 2
# Delta
device /dev/sdd1
raid-disk 3
[the partitions are 'fd' type: Linux auto raid]
After I created the physical volume with 'pvcreate /dev/md0' I got this
output with pvdisplay:
ludivinus:~# pvdisplay
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ --- Physical volume ---
~ PV Name /dev/cciss/c1d0p4
~ VG Name TequilaData
~ PV Size 60.15 GB / not usable 0
~ Allocatable yes
~ PE Size (KByte) 4096
~ Total PE 15398
~ Free PE 38
~ Allocated PE 15360
~ PV UUID eqEPJ1-YbVY-qhFQ-Fm63-9cC1-0UwZ-PAssTp
~ --- NEW Physical volume ---
~ PV Name /dev/md0
~ VG Name
~ PV Size 457.99 GB
~ Allocatable NO
~ PE Size (KByte) 0
~ Total PE 0
~ Free PE 0
~ Allocated PE 0
~ PV UUID jlq1y1-YpjZ-AU3Z-Iibp-0EkG-DzZj-5IHGhq
~ --- NEW Physical volume ---
~ PV Name /dev/sdd1
~ VG Name
~ PV Size 457.99 GB
~ Allocatable NO
~ PE Size (KByte) 0
~ Total PE 0
~ Free PE 0
~ Allocated PE 0
~ PV UUID *U|-?c-?6??I-?n??j-5IHGhq
its kinda strange, so I removed /dev/sdd1.
ludivinus:~# pvremove /dev/sdd1
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Labels on physical volume "/dev/sdd1" successfully wiped
and after that I pvdisplay showed the correct output. But after I
created the volume group and the logical volume I had this "ghost
volume" back again. So I removed it again. my initial tests look good.
there is no problem.
With my first test on this box, I created a raid just with the bare
disks without partitions, I had other problems, that he suddenly had a
partition in sda and sdd with the size of the right which screwed lvm
totaly up.
Anyway my question is, did I do something wrong? Missed I something out?
I couldn't find any good description for lvm on raid5. The only one I
could find, didn't sepcify anything else.
- --
Clemens Schwaighofer - IT Engineer & System Administration
==========================================================
Tequila Japan, 6-17-2 Ginza Chuo-ku, Tokyo 104-8167, JAPAN
Tel: +81-(0)3-3545-7703 Fax: +81-(0)3-3545-7343
http://www.tequila.jp
==========================================================
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
iD8DBQFAQrY9jBz/yQjBxz8RAvSFAKCKy9Vgzr3ZMSHkcY35d5MS9f8YGACfRQYu
rqZlE7SnIYii2w9QukfQLrk=
=xtAZ
-----END PGP SIGNATURE-----
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [linux-lvm] lvm on software raid 5
@ 2004-03-18 14:15 Rick van der Linde
2004-03-19 2:41 ` Ken Fuchs
0 siblings, 1 reply; 3+ messages in thread
From: Rick van der Linde @ 2004-03-18 14:15 UTC (permalink / raw)
To: linux-lvm
Hi there,
As I read your comments I would suggest you not to create an VG straight
onto the /dev/md0 device but first to create one huge partition on device
/dev/md0 with label 8e (Linux LVM) and create a VG on top of that
partition.
As I did something like this before I would like to know your experiences.
I've (succesfully) tried to install LVM on top of a RAID 5 volume
consisting 4 disks of 80GB. However I switched back to simple ext3 volumes
on RAID 5 duie to performance problems. I've tried this on LVM 1.0.3 and
the md-driver delivered with the 2.4.20 kernel. In this situation I
experience a (LVM on RAID5) perfomance loss of approx 40-60% in comparison
to the RAID5 perfomance. The performance advantages were more important to
me then the LVM flexibility was. I would like to know from you if the new
IO scheduler which comes with Kernel 2.6.X doesn't give a performance loss
anymore. (comparison true RAID5 perfomance to LVM on RAID5 performance) If
the performance hit is a few percent which is usual to expec from LVM this
will give me the optimal solution having a great performance AND maximum
flexibility. Thanks in advance....
Vriendelijke groeten/ Kindly regards,
Rick van der Linde
F u cn rd ths, u cn gt a gd jb n cmptr prgrmmng.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
hi,
I have a software raid 5 in linud 2.6.3 and I created a lvm on. Actually
it works, but there some strange things. Perhaps someone can point me
out why.
Basic:
lvm vesion:
ii lvm2 2.00.08-4 The Linux Logical Volume Manager
ii lvm-common 1.5.12 The Logical Volume Manager for
Linux (common files
raid version:
ii raidtools2 1.00.3-5 Utilities to support 'new-style'
RAID disks
md admin version
ii mdadm 1.4.0-3 Manage MD devices aka Linux
Software Raid
my raidtab:
# 4 x 160GB Raid 5, no spare
raiddev /dev/md0
raid-level 5
nr-raid-disks 4
nr-spare-disks 0
persistent-superblock 1
chunk-size 64
parity-algorithm left-symmetric
# Spare disks for hot reconstruction
# no the raid devices
# Alpha
device /dev/sda1
raid-disk 0
# Beta
device /dev/sdb1
raid-disk 1
# Gamma
device /dev/sdc1
raid-disk 2
# Delta
device /dev/sdd1
raid-disk 3
[the partitions are 'fd' type: Linux auto raid]
After I created the physical volume with 'pvcreate /dev/md0' I got this
output with pvdisplay:
ludivinus:~# pvdisplay
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ --- Physical volume ---
~ PV Name /dev/cciss/c1d0p4
~ VG Name TequilaData
~ PV Size 60.15 GB / not usable 0
~ Allocatable yes
~ PE Size (KByte) 4096
~ Total PE 15398
~ Free PE 38
~ Allocated PE 15360
~ PV UUID eqEPJ1-YbVY-qhFQ-Fm63-9cC1-0UwZ-PAssTp
~ --- NEW Physical volume ---
~ PV Name /dev/md0
~ VG Name
~ PV Size 457.99 GB
~ Allocatable NO
~ PE Size (KByte) 0
~ Total PE 0
~ Free PE 0
~ Allocated PE 0
~ PV UUID jlq1y1-YpjZ-AU3Z-Iibp-0EkG-DzZj-5IHGhq
~ --- NEW Physical volume ---
~ PV Name /dev/sdd1
~ VG Name
~ PV Size 457.99 GB
~ Allocatable NO
~ PE Size (KByte) 0
~ Total PE 0
~ Free PE 0
~ Allocated PE 0
~ PV UUID *U|-?c-?6??I-?n??j-5IHGhq
its kinda strange, so I removed /dev/sdd1.
ludivinus:~# pvremove /dev/sdd1
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Incorrect metadata area header checksum
~ Labels on physical volume "/dev/sdd1" successfully wiped
and after that I pvdisplay showed the correct output. But after I
created the volume group and the logical volume I had this "ghost
volume" back again. So I removed it again. my initial tests look good.
there is no problem.
With my first test on this box, I created a raid just with the bare
disks without partitions, I had other problems, that he suddenly had a
partition in sda and sdd with the size of the right which screwed lvm
totaly up.
Anyway my question is, did I do something wrong? Missed I something out?
I couldn't find any good description for lvm on raid5. The only one I
could find, didn't sepcify anything else.
- --
Clemens Schwaighofer - IT Engineer & System Administration
==========================================================
Tequila Japan, 6-17-2 Ginza Chuo-ku, Tokyo 104-8167, JAPAN
Tel: +81-(0)3-3545-7703 Fax: +81-(0)3-3545-7343
http://www.tequila.jp
==========================================================
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
iD8DBQFAQrY9jBz/yQjBxz8RAvSFAKCKy9Vgzr3ZMSHkcY35d5MS9f8YGACfRQYu
rqZlE7SnIYii2w9QukfQLrk=
=xtAZ
-----END PGP SIGNATURE-----
_______________________________________________
linux-lvm mailing list
linux-lvm@redhat.com
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [linux-lvm] lvm on software raid 5
2004-03-18 14:15 Rick van der Linde
@ 2004-03-19 2:41 ` Ken Fuchs
0 siblings, 0 replies; 3+ messages in thread
From: Ken Fuchs @ 2004-03-19 2:41 UTC (permalink / raw)
To: linux-lvm
Rick van der Linde wrote:
>As I did something like this before I would like to know your experiences.
>I've (successfully) tried to install LVM on top of a RAID 5 volume
>consisting 4 disks of 80GB. However I switched back to simple ext3 volumes
>on RAID 5 due to performance problems. I've tried this on LVM 1.0.3 and
>the md-driver delivered with the 2.4.20 kernel. In this situation I
>experience a (LVM on RAID5) performance loss of approx 40-60% in comparison
>to the RAID5 performance.
That is a huge performance loss. Usually the LVM extents will be
sequentially ordered within each logical volume providing good
performance. Don't forget that the raw performance of the outer
cylinders of a drive can be double that of the inner cylinders. Any
performance test setup that doesn't neutralize this variable has no
value period. I'm not suggesting that this is the case, but only that
performance tests must be very carefully designed and executed.
I did some tests on LVM 1.0.8 (no RAID) using bonnie and noticed only a
few percent performance drop.
Sincerely,
Ken Fuchs <kfuchs@winternet.com>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2004-03-19 2:42 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-02-29 23:03 [linux-lvm] lvm on software raid 5 Clemens Schwaighofer
-- strict thread matches above, loose matches on Subject: below --
2004-03-18 14:15 Rick van der Linde
2004-03-19 2:41 ` Ken Fuchs
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox