Hi, I have a machine booting off IDE RAID. /dev/md0 / reiserfs defaults 1 1 /dev/hda2 swap swap pri=42 0 0 /dev/hdc2 swap swap pri=42 0 0 What I'm wondering is can I convert my dual swaps to one single RAIDed swap? Dave.
On Thursday 18 March 2004 14:46, Dave Lists wrote:
Hi, I have a machine booting off IDE RAID. /dev/md0 / reiserfs defaults 1 1 /dev/hda2 swap swap pri=42 0 0 /dev/hdc2 swap swap pri=42 0 0
What I'm wondering is can I convert my dual swaps to one single RAIDed swap?
Dave.
Why would you want to do that? Raid is slower than regular ide because of the need to sync the drives. Raiding your swaps will simply slow you down. Before you waste any time on this, evalate just how much of the time you are useing swap. With just a half meg of memory and vmware running my system still shows virtally no swap useage at all: jsa@pen:~> free total used free shared buffers cached Mem: 514152 507196 6956 0 28532 187704 -/+ buffers/cache: 290960 223192 Swap: 1028120 4936 1023184 Swap does not have to be persistant so why waste the cycles raiding it up? -- _____________________________________ John Andersen
If you lose your swap drive during machine uptime, most likely the machine will crash. Hence one benefit of putting swap on RAID is increased fault tolerance, although with software RAID you do have to balance that with a slight decrease in performance. John Andersen wrote:
On Thursday 18 March 2004 14:46, Dave Lists wrote:
Hi, I have a machine booting off IDE RAID. /dev/md0 / reiserfs defaults 1 1 /dev/hda2 swap swap pri=42 0 0 /dev/hdc2 swap swap pri=42 0 0
What I'm wondering is can I convert my dual swaps to one single RAIDed swap?
Dave.
Why would you want to do that?
Raid is slower than regular ide because of the need to sync the drives. Raiding your swaps will simply slow you down.
Before you waste any time on this, evalate just how much of the time you are useing swap.
With just a half meg of memory and vmware running my system still shows virtally no swap useage at all: jsa@pen:~> free total used free shared buffers cached Mem: 514152 507196 6956 0 28532 187704 -/+ buffers/cache: 290960 223192 Swap: 1028120 4936 1023184
Swap does not have to be persistant so why waste the cycles raiding it up?
-- --Moby They that can give up essential liberty to obtain a little temporary safety deserve neither liberty nor safety. -- Benjamin Franklin First they came for the Jews and I did not speak out because I was not a Jew. Then they came for the Communists and I did not speak out because I was not a Communist. Then they came for the trade unionists and I did not speak out because I was not a trade unionist. Then they came for me and there was no one left to speak out for me. -- Pastor Martin Niemöller
On Thursday 18 March 2004 18:42, Mobeen Azhar wrote:
If you lose your swap drive during machine uptime, most likely the machine will crash. Hence one benefit of putting swap on RAID is increased fault tolerance, although with software RAID you do have to balance that with a slight decrease in performance.
I have lost swap drives in the past. Ran half a week without noticing. Some processes died when they could not swap stuff back IN. But the machine ran fine since (as is normally the case) swap was used very little. If one insists on mirroring swap its best to get one of those hardware raid controllers which eliminated the need for software raid and often the OS is totally unaware it has a raid. Those things are rally cheap these days ($30) if all you want is raid 1 mirroring. -- _____________________________________ John Andersen
John Andersen wrote:
If one insists on mirroring swap its best to get one of those hardware raid controllers which eliminated the need for software raid and often the OS is totally unaware it has a raid. Those things are rally cheap these days ($30) if all you want is raid 1 mirroring.
Those cheap "hardware" RAID controllers are actually semi-software controllers - they rely on the OS's RAID implementation for most of the actual work. If you want true hardware RAID then expect to spend a couple hundred dollars. The 3ware ones seem pretty decent. So having said that, relying on the Linux software RAID implementation does't seem like such a bad idea.
On Thursday 18 March 2004 20:28, Avtar Gill wrote:
John Andersen wrote:
If one insists on mirroring swap its best to get one of those hardware raid controllers which eliminated the need for software raid and often the OS is totally unaware it has a raid. Those things are rally cheap these days ($30) if all you want is raid 1 mirroring.
Those cheap "hardware" RAID controllers are actually semi-software controllers - they rely on the OS's RAID implementation for most of the actual work. If you want true hardware RAID then expect to spend a couple hundred dollars. The 3ware ones seem pretty decent. So having said that, relying on the Linux software RAID implementation does't seem like such a bad idea.
You can tell by the chipset if its a real raid controller, also if it has raid setup software onboard. (And yes, still at $30). If it has a utility for defining raid arrays chances are you system will not even see separate drives, in which case there is no OS involvement. I've even moved one of these drive arrays originally set up on Windows to Linux and it mounted it, found the vfat partition and was none the wiser. You don't have to spend that kind of money any more unless you want high-performance raid 5. (Oh, and the thing performed deciently too...) -- _____________________________________ John Andersen
John Andersen
You can tell by the chipset if its a real raid controller,
This may be so.
also if it has raid setup software onboard. (And yes, still at $30).
But this not! Even the cheap Promise Fasttracks have RAID setup code in the BIOS but the rest is done in the driver.
If it has a utility for defining raid arrays chances are you system will not even see separate drives, in which case there is no OS involvement.
Not necessarily, it could also mean that the chip uses a trick to hide the additional disk(s) like Promise does in its fake RAID cards. The Linux kernel can make them visible again.
I've even moved one of these drive arrays originally set up on Windows to Linux and it mounted it, found the vfat partition and was none the wiser.
No proof. That's exactly what ataraid and pdcraid/hptraid are there for: to translate the disk setup and make it understandable for the md (softraid) driver.
You don't have to spend that kind of money any more unless you want high-performance raid 5.
I still don't believe it. All real hardware IDE RAID controllers I've seen cost *way* beyond those $ 30. -- Philipp Thomas work: pth AT suse DOT de SUSE LINUX AG private: philipp DOT thomas AT t-link DOT de
On Sat, 2004-03-20 at 16:14, Philipp Thomas wrote:
I still don't believe it. All real hardware IDE RAID controllers I've seen cost *way* beyond those $ 30.
Here's a server that someone let me use that *only* SuSE 9.0 was able to be setup on. We tried RH 7-9, FBSD 4-5.2, Debian, etc. I knew SuSE would work :-) SCSI subsystem driver Revision: 1.00 kmod: failed to exec /sbin/modprobe -s -k scsi_hostadapter, errno = 2 Loading Adaptec I2O RAID: Version 2.4 Build 5 Detecting Adaptec I2O RAID controllers... Adaptec I2O RAID controller 0 at f8dfb000 size=100000 irq=16 dpti: If you have a lot of devices this could take a few minutes. dpti0: Reading the hardware resource table. TID 008 Vendor: ADAPTEC Device: AIC-7899 Rev: 00000001 TID 519 Vendor: ADAPTEC Device: RAID-5 Rev: 370F scsi0 : Vendor: Adaptec Model: 2100S FW:370F Vendor: ADAPTEC Model: RAID-5 Rev: 370F Type: Direct-Access ANSI SCSI revision: 02 Attached scsi disk sda at scsi0, channel 0, id 0, lun 0 Dee
participants (6)
-
Avtar Gill
-
Dave Lists
-
John Andersen
-
Mobeen Azhar
-
Philipp Thomas
-
W.D.McKinney