RAID 5 questions
Finally resurrected my home server and got RAID 5 on it working, got few questions about the array itself, but first the config:
Mobo: Supermicro P3TDLR+
CPUs: 2x Pentium III Tualatin 1,4GHz
RAM: 2x1GB PC133 ECC / buffered
Storage (IDE) : 1x HP DVD-ROM (PATA)
1x IBM DeathStar 80GB (PATA)
(SATA): PCI-X FastTrak x4300 SATAII Hot Swappable
4x Seagate 1TB in RAID5 (SATAII)
And the last one :
More details here : http://www.supermicro.nl/products/ch...C733TQ-645.cfm
Setting up the system (with XP SP2 32bit) was easy, tweaks and stuff done, ALL OK. What bothers me is the array:
RAID 5 gave me a total of 3TB of space and 1TB for the parity - that was expected. Placed 128KB block mode, write-back cache processing, created one 2TB partition and one 1TB partition, since 32bit OS supports only 2TB partitions (but I think some othe OSes can go higher on 32bit somehow.), formatted and copied some .avi files on it from my IDE HDD to perform this test:
Wanted to see the RAID5 3+1 redundancy, so during the copy back of those avis to the IDE drive I unplugged one of those drives - ALL OK, copy back was still running, so I unplugged a second one, which displayed the read error - ALL OK. Plugged the drives back, cancelled the errors and tried copying once more - this time it was stragne, at least I understand it like this (and maybe wrongly): The array was back online, but when I wanted to copy again, I got only 3 drives working, but the copying still worked. When restarted, the BIOS of the controller prompted me for action, because one of the drive could not be found or failed - I thought that was the part of rebuilding my array with parity's assistance, but there was no such option that could allow me to do so - I only had the information table and the capability to delete (thus losing data according to the warning table) and create the logical devices. First I thought the drive really failed, but after deleting the LDEV and creating a new one, the same array, same attributes, everything seemed fine, the drive was back online and arranged into the array. What really bothers me that the .avi files were still there
!!!
WHAT? I thought the data are lost when I try to delete and create a new LDEV, no matter if I change the attributes or not. Please clarify.
Also I'd like to ask if there is any logic in RAID failures, for instance I have my 0-3 drives, and let's say I wanna take out the 2nd and the 4th drive (1 & 3), which one of these should NOT be working, or ''ungroup'' itself? Because I have a feeling that when I tried particularly remove the drives no. 1 & 2 and in that order, the no. 1 failed, while the no. 2 worked with the rest of the drives - shouldn't that be vice versa? Or this happens only when I write to the array. Really I'm out of knowledge when grouping / ungrouping drives etc. Should I use some specific software for this type of procedure?
Mobo: Supermicro P3TDLR+
CPUs: 2x Pentium III Tualatin 1,4GHz
RAM: 2x1GB PC133 ECC / buffered
Storage (IDE) : 1x HP DVD-ROM (PATA)
1x IBM DeathStar 80GB (PATA)
(SATA): PCI-X FastTrak x4300 SATAII Hot Swappable
4x Seagate 1TB in RAID5 (SATAII)
And the last one :
More details here : http://www.supermicro.nl/products/ch...C733TQ-645.cfm
Setting up the system (with XP SP2 32bit) was easy, tweaks and stuff done, ALL OK. What bothers me is the array:
RAID 5 gave me a total of 3TB of space and 1TB for the parity - that was expected. Placed 128KB block mode, write-back cache processing, created one 2TB partition and one 1TB partition, since 32bit OS supports only 2TB partitions (but I think some othe OSes can go higher on 32bit somehow.), formatted and copied some .avi files on it from my IDE HDD to perform this test:
Wanted to see the RAID5 3+1 redundancy, so during the copy back of those avis to the IDE drive I unplugged one of those drives - ALL OK, copy back was still running, so I unplugged a second one, which displayed the read error - ALL OK. Plugged the drives back, cancelled the errors and tried copying once more - this time it was stragne, at least I understand it like this (and maybe wrongly): The array was back online, but when I wanted to copy again, I got only 3 drives working, but the copying still worked. When restarted, the BIOS of the controller prompted me for action, because one of the drive could not be found or failed - I thought that was the part of rebuilding my array with parity's assistance, but there was no such option that could allow me to do so - I only had the information table and the capability to delete (thus losing data according to the warning table) and create the logical devices. First I thought the drive really failed, but after deleting the LDEV and creating a new one, the same array, same attributes, everything seemed fine, the drive was back online and arranged into the array. What really bothers me that the .avi files were still there
!!!WHAT? I thought the data are lost when I try to delete and create a new LDEV, no matter if I change the attributes or not. Please clarify.
Also I'd like to ask if there is any logic in RAID failures, for instance I have my 0-3 drives, and let's say I wanna take out the 2nd and the 4th drive (1 & 3), which one of these should NOT be working, or ''ungroup'' itself? Because I have a feeling that when I tried particularly remove the drives no. 1 & 2 and in that order, the no. 1 failed, while the no. 2 worked with the rest of the drives - shouldn't that be vice versa? Or this happens only when I write to the array. Really I'm out of knowledge when grouping / ungrouping drives etc. Should I use some specific software for this type of procedure?
I hate hardware RAID. Every experience I've ever had with it was a disaster. Software RAID has served me well in every flavor I've ever tried. mdadm, MS Storage Spaces, StableBit DrivePool, Intel fakeraid, etc. So, for my "new" VM server at home, running Proxmox, I decided to go for ZFS.
Comment