RAID Hardware Question
- Thread starter mcit
- Start date
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
If no one knows the answer to that, then alternativley, I have an Adaptec 4805 SAS card that no matter how I build the array, proxmox will not install on, it gets all the way to creating the partitions then failes with a line 547 error. I can install the latest Debian build on the same array without a problem though. Does anyone have any experience with one of these cards?
Hi mcit,
I am using a Highpoint 3510 and it works very well with proxmox (version 1.5, .32 Kernel). But looking at the Highpoints site the 4320 is supported by Debian 5. So I think this card will work too.
Can you not test order it and try it?
Kind regards
I am using a Highpoint 3510 and it works very well with proxmox (version 1.5, .32 Kernel). But looking at the Highpoints site the 4320 is supported by Debian 5. So I think this card will work too.
Can you not test order it and try it?
Kind regards
D
Dragonchaser
Guest
I don't know about the Highpoint but I did read that 3Ware works fine:
02-17-2009 #22 tom
just to add my 2 cents:
hard disks also include cache memory which is not protected by batteries backup.
so you need to disable the hard drive cache.
therefore, if you have soft raid, you have no hard drive write cache at all - bad
performance.
if you have hardware raid you can use the batteries protected cache from the
controller (do not forget to disable the cache from the hard disks).
consumer hard disk has the cache enabled, server disk sometimes too. so if you
want a stable and fast system, pls double check the cache settings.
I assume a lot of server problems around (not only on Proxmox VE, also in the
windows world) are caused by not protected disk cache memory and powerloss.
best regards,
tom
05-13-2009 #38 gijsbert
I fully agree with lborkey, Proxmox VE is almost perfect, and the inclusion of
software-raid would make Proxmox VE the perfect solution. From a development point
of view however, I can understand that proxmox will not support software-raid. So
first of all, proxmox: great work!!!
I think most of the people would like to see a software raid-1 solution as some
kind of "basic protection". For other raid-levels (and performance) they should
buy a hardware-raid card. By the way, 3ware 9650 and 9690 serie works with proxmox
Why do I need a software raid-1 solution: We manage > 100 servers and we like to
add a openvz virtualization layer on each server (1 container per server). With
this setup it's much easier in the future to do migrations and upgrades. All these
servers currently run centos5 with software raid-1. Adding a good hardware
raid-controller + BBU on each server will cost approx. 50.000 euro and
unfortenately that's not an option That's the reason (for us) why software raid-1
would be perfect.
I like to stich with proxmox because the graphical interface is very well done,
but without software raid-1 setup it's a no go. Does anyone know a webbased
management tool for openvz which does support software raid-1?
Gijsbert
05-14-2009 #45
We're currently testing performance (read/write) of disks running CentOS and
different virtual environments (proxmox and xenserver) in a soft- and
hardware-raid-1 (3ware 9690). If you're interested in the results, please let me
know.
(copy pasted from another thread, hope this helps)
02-17-2009 #22 tom
just to add my 2 cents:
hard disks also include cache memory which is not protected by batteries backup.
so you need to disable the hard drive cache.
therefore, if you have soft raid, you have no hard drive write cache at all - bad
performance.
if you have hardware raid you can use the batteries protected cache from the
controller (do not forget to disable the cache from the hard disks).
consumer hard disk has the cache enabled, server disk sometimes too. so if you
want a stable and fast system, pls double check the cache settings.
I assume a lot of server problems around (not only on Proxmox VE, also in the
windows world) are caused by not protected disk cache memory and powerloss.
best regards,
tom
05-13-2009 #38 gijsbert
I fully agree with lborkey, Proxmox VE is almost perfect, and the inclusion of
software-raid would make Proxmox VE the perfect solution. From a development point
of view however, I can understand that proxmox will not support software-raid. So
first of all, proxmox: great work!!!
I think most of the people would like to see a software raid-1 solution as some
kind of "basic protection". For other raid-levels (and performance) they should
buy a hardware-raid card. By the way, 3ware 9650 and 9690 serie works with proxmox
Why do I need a software raid-1 solution: We manage > 100 servers and we like to
add a openvz virtualization layer on each server (1 container per server). With
this setup it's much easier in the future to do migrations and upgrades. All these
servers currently run centos5 with software raid-1. Adding a good hardware
raid-controller + BBU on each server will cost approx. 50.000 euro and
unfortenately that's not an option That's the reason (for us) why software raid-1
would be perfect.
I like to stich with proxmox because the graphical interface is very well done,
but without software raid-1 setup it's a no go. Does anyone know a webbased
management tool for openvz which does support software raid-1?
Gijsbert
05-14-2009 #45
We're currently testing performance (read/write) of disks running CentOS and
different virtual environments (proxmox and xenserver) in a soft- and
hardware-raid-1 (3ware 9690). If you're interested in the results, please let me
know.
(copy pasted from another thread, hope this helps)
Ok, so I managed to get a Highpoint 4320 card for testing. I create the Array [6x 1.5Tb SATA in RAID5], then try to install.
Same error. Exact error is below. Can someone tell me what this means and how to fix it. When I get this error I am unable to press Alt+F2 for more information, nothing comes up on the screen. I can only Abort or Reboot.
"Unable to create data volume at /usr/bin/proxinstall line 567"
Same error. Exact error is below. Can someone tell me what this means and how to fix it. When I get this error I am unable to press Alt+F2 for more information, nothing comes up on the screen. I can only Abort or Reboot.
"Unable to create data volume at /usr/bin/proxinstall line 567"
Hi,Ok, so I managed to get a Highpoint 4320 card for testing. I create the Array [6x 1.5Tb SATA in RAID5], then try to install.
Same error. Exact error is below. Can someone tell me what this means and how to fix it. When I get this error I am unable to press Alt+F2 for more information, nothing comes up on the screen. I can only Abort or Reboot.
"Unable to create data volume at /usr/bin/proxinstall line 567"
your volume for the installation (raid-partition) should be less than 2 TB. Is it so?
Udo
Ok, I have solved most of my problems with the help of this thread.
http://forum.proxmox.com/threads/1151-Install-on-single-large-Raid-Lun-(-3.8Tb)
But, I would like to use 6TB. The problem is that I seem to hit a wall at the 4tb mark. Does anyone know how I can get past this? I seem to be unable to create any more LVs as proxmox CLI does not see any more space.
http://forum.proxmox.com/threads/1151-Install-on-single-large-Raid-Lun-(-3.8Tb)
But, I would like to use 6TB. The problem is that I seem to hit a wall at the 4tb mark. Does anyone know how I can get past this? I seem to be unable to create any more LVs as proxmox CLI does not see any more space.
Ok. I have solved my problem. For the sake of finishing off the thread here. This is what I have done. Please note this is not an exact set of steps, just an overall indication of what got me there.
Install proxmox on a mirror 80gb arrary [seperate HDD's in my case]
Install parted
apt-get update
apt-get install parted
Use parted to create 1 single large 5.6TB partition [6x 1.5TB HDDs in RAID5]
This became /dev/sdb1
pvcreate /dev/sdb1
vgextend pve /dev/sdb1
lvextend /dev/pve/data -L +5600G
Please note that I might have some of these commands slightly off, so check before copying them.
Install proxmox on a mirror 80gb arrary [seperate HDD's in my case]
Install parted
apt-get update
apt-get install parted
Use parted to create 1 single large 5.6TB partition [6x 1.5TB HDDs in RAID5]
This became /dev/sdb1
pvcreate /dev/sdb1
vgextend pve /dev/sdb1
lvextend /dev/pve/data -L +5600G
Please note that I might have some of these commands slightly off, so check before copying them.
Hi,...
pvcreate /dev/sdb1
vgextend pve /dev/sdb1
lvextend /dev/pve/data -L +5600G
...
but in this case your logical volume data are on both arrays - the "slow" mirror and the hopefully fast array.
To use only sdb1 for data you can:
1. backup /var/lib/vz; remove lv data; create data new as continous lv (-C y) - it's perhaps a problem if you wan't extend the lv later.
2. normal install on sda (mirror); backup /var/lib/vz; remove lv data; create dummy lv to fill pve completely; extend pve with sdb1; create data on new space; remove dummy lv (this space can be used for a backup-lv and free space for snapshots (backup))
After that you must of course create an new ext3-filesystem on data and restore the backup...
Udo
This is an update for anyone who would like the steps to do this properly [I say properly as I believe this is the correct way for this to be done]
Proxmox Create Volume Group Data >2TB
lvdisplay –v [Show current logical volumes]
umount /dev/pve/data [unmount logical volume 'data'
lvremove /dev/pve/data [ensure you backup anything there first to another location or server]
pvdisplay –v [show current physical volumes]
pvcreate /dev/sdxxx [create new physical volume for use. This is your new HDD or array. If you are unsure the exact device, look under /dev dir. You can also use fdisk -l to help]
pvdisplay –v [Your new physical volume should now be visable]
vgdisplay –v [display size and config of volume group]
vgextend /dev/pve /dev/sdxxx [This will extend the volume group by the new physical volume you have added above]
lvcreate –n /dev/pve/data –L sizeG [This will create your new /data store in the volume group Note: The G after size is a value in Gigabytes]
lvdisplay –v [You should now see your new data store with the larger size, this can be >2TB]
mkfs –T ext3 /dev/pve/data [This will format your new store ready for use]
You do not need to reboot, but it is a good idea to ensure you have not broken your server.
I have used this method on several servers of various array sizes between 2TB and 8TB.
Proxmox Create Volume Group Data >2TB
lvdisplay –v [Show current logical volumes]
umount /dev/pve/data [unmount logical volume 'data'
lvremove /dev/pve/data [ensure you backup anything there first to another location or server]
pvdisplay –v [show current physical volumes]
pvcreate /dev/sdxxx [create new physical volume for use. This is your new HDD or array. If you are unsure the exact device, look under /dev dir. You can also use fdisk -l to help]
pvdisplay –v [Your new physical volume should now be visable]
vgdisplay –v [display size and config of volume group]
vgextend /dev/pve /dev/sdxxx [This will extend the volume group by the new physical volume you have added above]
lvcreate –n /dev/pve/data –L sizeG [This will create your new /data store in the volume group Note: The G after size is a value in Gigabytes]
lvdisplay –v [You should now see your new data store with the larger size, this can be >2TB]
mkfs –T ext3 /dev/pve/data [This will format your new store ready for use]
You do not need to reboot, but it is a good idea to ensure you have not broken your server.
I have used this method on several servers of various array sizes between 2TB and 8TB.
Hi,This is an update for anyone who would like the steps to do this properly [I say properly as I believe this is the correct way for this to be done]
Proxmox Create Volume Group Data >2TB
lvdisplay –v [Show current logical volumes]
umount /dev/pve/data [unmount logical volume 'data'
lvremove /dev/pve/data [ensure you backup anything there first to another location or server]
pvdisplay –v [show current physical volumes]
pvcreate /dev/sdxxx [create new physical volume for use. This is your new HDD or array. If you are unsure the exact device, look under /dev dir. You can also use fdisk -l to help]
pvdisplay –v [Your new physical volume should now be visable]
vgdisplay –v [display size and config of volume group]
vgextend /dev/pve /dev/sdxxx [This will extend the volume group by the new physical volume you have added above]
lvcreate –n /dev/pve/data –L sizeG [This will create your new /data store in the volume group Note: The G after size is a value in Gigabytes]
lvdisplay –v [You should now see your new data store with the larger size, this can be >2TB]
mkfs –T ext3 /dev/pve/data [This will format your new store ready for use]
You do not need to reboot, but it is a good idea to ensure you have not broken your server.
I have used this method on several servers of various array sizes between 2TB and 8TB.
in this case, you don't need to remove the lv (vgetend and lvextend are enough) - because the lv-data go over both devices!! If you don't want that, fill first (before extend pve) the pve-vg with an dummy-lv.
Udo