raictl - ?
ki_chi_saga
I'mt trying to setup a raid1 using raidctl but despite the disks are
identical I receive;
.........................................................................................................................................
# raidctl -c c1t0d0 c1t1d0
Creating RAID volume will destroy all data on spare space of member
disks, proceed (yes/no)? yes
Disk has occupied space
.........................................................................................................................................
# prtvtoc -s /dev/rdsk/c1t1d0s2
* First Sector Last
* Partition Tag Flags Sector Count Sector Mount Directory
0 2 00 0 286248144 286248143
2 5 00 0 286248144 286248143
.............................................................................
# prtvtoc -s /dev/rdsk/c1t0d0s2
* First Sector Last
* Partition Tag Flags Sector Count Sector Mount Directory
0 2 00 0 286248144 286248143
2 5 00 0 286248144 286248143
.............................................................................
0. c1t0d0 <LSILOGIC-LogicalVolume-3000 cyl 65533 alt 2 hd 16 sec
273>
/pci@0/pci@0/pci@2/LSILogic,sas@0/sd@0,0
1. c1t1d0 <LSILOGIC-LogicalVolume-3000 cyl 65533 alt 2 hd 16 sec 273
/pci@0/pci@0/pci@2/LSILogic,sas@0/sd@1,0
..........................................................................................................
Both disk contains only a zfs rpool.
Solaris is a newinstalled Solaris 10.
Anybody know why?
Rgds,
Mat
John D Groenveld
ki_chi_saga <fan...@kth.se> wrote:
>I'mt trying to setup a raid1 using raidctl but despite the disks are
>identical I receive;
>.........................................................................................................................................
># raidctl -c c1t0d0 c1t1d0
>Creating RAID volume will destroy all data on spare space of member
>disks, proceed (yes/no)? yes
>Disk has occupied space
>Both disk contains only a zfs rpool.
>Solaris is a newinstalled Solaris 10.
Why did you decide against ZFS mirroring?
John
groe...@acm.org
Doug McIntyre
>I'mt trying to setup a raid1 using raidctl but despite the disks are
>identical I receive;
>Creating RAID volume will destroy all data on spare space of member
>disks, proceed (yes/no)? yes
>Disk has occupied space
I assume this is the error you are getting, but then you state..
>Both disk contains only a zfs rpool.
The util tells you the disks have data on them and won't do anything
(without presuming you whack it all and start again), and you state
that there is data in the form of ZFS pools on it.
So I don't see the issue? Seems to be working correctly to help
you not whack the system away underneath your feet.
raidctl is used at the beginning before there is any data on the
disks, and massages the hardware RAID cards in a system. Not after
there is already a system installed on it. You can't mirror drives
after the fact, they have to be setup before anything is there.
ZFS prefers working directly with the disks without any hardware RAID
in the way. Solaris gives you many options about how you'd want to do
thing, and since the system has the hardware RAID card, some people
insist on using it. But I'd say the recommended setup is to forgo
using the hadware RAID and just go ZFS (presuming that this is a
mirror pool for ZFS) and you're all set.
ki_chi_saga
> In article <285b76a4-328a-40bd-ac06-e151659e5...@28g2000yqu.googlegroups.com>,
>
> Why did you decide against ZFS mirroring?
>
> John
We're doing this to compare performance for build-jobs in a Sun Grid
Engine cluster.
Servers in the cluster will be migrated to zfs, and previous
experinces, from others
and from within my own organisation, tells us that we might end up
with jobs having
longer execution times having mirroring setup within zfs than using HW-
raid.
Rgds,
Mats
ki_chi_saga
>
> I assume this is the error you are getting, but then you state..
>
> >Both disk contains only a zfs rpool.
>
> The util tells you the disks have data on them and won't do anything
> (without presuming you whack it all and start again), and you state
> that there is data in the form of ZFS pools on it.
>
> So I don't see the issue? Seems to be working correctly to help
> you not whack the system away underneath your feet.
>
> raidctl is used at the beginning before there is any data on the
> disks, and massages the hardware RAID cards in a system. Not after
> there is already a system installed on it. You can't mirror drives
> after the fact, they have to be setup before anything is there.
>
Thanks for your thorough response.
>> So I don't see the issue?
After having read what you wrote,
neigher do I ....
Thanks for opening my eys .....
Rgds,
Mat
John D Groenveld
ki_chi_saga <fan...@kth.se> wrote:
>We're doing this to compare performance for build-jobs in a Sun Grid
>Engine cluster.
>Servers in the cluster will be migrated to zfs, and previous
>experinces, from others
>and from within my own organisation, tells us that we might end up
>with jobs having
>longer execution times having mirroring setup within zfs than using HW-
>raid.
Since you're reinstalling, why not configure the HW RAID from LSI's
BIOS?
Alternatively you can probably overwrite disk label with dd(1) and
then run raidctl(1M) to configure the HW RAID.
Let us know what if any performance difference you see between
the two configurations.
John
groe...@acm.org
cindy
I hope its clear that the ZFS root pool can only be a single disk or a
mirrored pool of x number of disks. An existing limitation is that th
boot information must be contiguous. You couldn't use two separate
disks
for a root pool unless you glommed them together as one LUN,
which I wouldn't recommend for a root pool.
You might consider using two smaller disks just for the root pool
and mirror them. Then, keep your data pools separate on HW-RAID.
You might be surprised, though, with mirrored ZFS pool performance,
which is also easy and flexible to manage.
Thanks,
Ciindy
ki_chi_saga
>
> Since you're reinstalling, why not configure the HW RAID from LSI's
> BIOS?
In our Sun Grid Engine cluster there is about 80 instances on 7
physical boxes, the rest is containers, so our first choice will be
to
migrate these to zfs.
But if I interpret Cindy's response below correct that would not be
possible
if we would want to use HW-raid.
Also, using raidctl would have been much better than configure the HW
RAID from the LSI's BIOS
because it's easier (an in my opinion better) to monitor the health of
the mirror from
within the host. That can, as far as I know, not be done if configured
from LSI's BIOS.
>
> Alternatively you can probably overwrite disk label with dd(1) and
> then run raidctl(1M) to configure the HW RAID.
>
After what you wrote earlier, I've completely wiped out all data on
the drive,
i.e reformat it and also deleted the entire partition, but all I get
is the
'Disk has occupied space' message.
Thanx for your response.
Rgds,
Mat
ki_chi_saga
> I hope its clear that the ZFS root pool can only be a single disk or a
> boot information must be contiguous. You couldn't use two separate
> disks for a root pool unless you glommed them together as one LUN,
> which I wouldn't recommend for a root pool.
This was news, to say the least ....
But explains it all ...
Thanx for your thorough response!
Rgds,
Mat
Scott
>
>
> > Alternatively you can probably overwrite disk label with dd(1) and
> > then run raidctl(1M) to configure the HW RAID.
>
> After what you wrote earlier, I've completely wiped out all data on
> the drive,
> i.e reformat it and also deleted the entire partition, but all I get
> is the
> 'Disk has occupied space' message.
>
> Thanx for your response.
# dd if=/dev/zero of=/dev/rdsk/c1t0d0s2 bs=512 count=100
# dd if=/dev/zero of=/dev/rdsk/c1t1d0s2 bs=512 count=100
I'd be somewhat surprised if they complain again.
Regards, Scott
Stefan Krueger
Didn't the orignal author say there were some zpools on the mentioned
disks? IIRC zfs writes an EFI label on them, which puts headers not
only at the beginning of the disk but also at the end
see http://en.wikipedia.org/wiki/GUID_Partition_Table
HTH
ki_chi_saga
100+0 records in
100+0 records out
# uname -a
SunOS selix063gh 5.10 Generic_141444-09 sun4v sparc SUNW,Sun-Blade-
T6320
# raidctl -c c1t0d0 c1t1d0
Creating RAID volume will destroy all data on spare space of member
disks, proceed (yes/no)? yes
Disk has occupied space.
The info Cindy wrote was a suprise to me as well ...,
but it apears she's right.
Rgds,
Mat
ki_chi_saga
> Didn't the orignal author say there were some zpools on the mentioned
> disks? IIRC zfs writes an EFI label on them, which puts headers not
> only at the beginning of the disk but also at the end
>
> seehttp://en.wikipedia.org/wiki/GUID_Partition_Table
>
> HTH
Maybe I expressed myself unclear, but there is only one zpool;
--------------------------------------------------------
zpool status
pool: rpool
state: ONLINE
scrub: none requested
config:
NAME STATE READ WRITE CKSUM
rpool ONLINE 0 0 0
c1t0d0s0 ONLINE 0 0 0
errors: No known data errors
.............................................................
#zfs list
NAME USED AVAIL REFER MOUNTPOINT
rpool 8.34G 126G 97K /rpool
rpool/ROOT 4.84G 126G 21K legacy
rpool/ROOT/s10s_u8wos_08a 4.84G 126G 4.84G /
rpool/dump 1.50G 126G 1.50G -
rpool/export 44K 126G 23K /export
rpool/export/home 21K 126G 21K /export/home
rpool/swap
--------------------------------------------------------
The drive c1t1do *is* for sure empty now (...),
but it doesn't make any difference,
all raidctl says is "Disk has occupied space."
Rgds,
Mat
Sami Ketola
> Maybe I expressed myself unclear, but there is only one zpool;
If you are using ZFS already on the first disk, don't even think
about using raidctl to mirror it. Use ZFS to mirror the pool.
zpool attach rpool c1t0d0s0 c1t1d0s0
Sami
cindy
As Mat said, this is rpool, the pool used for booting so it would not
have an EFI label. We can't currently boot from an EFI labeled disk.
Thanks,
Cindy