Recommended NAS?

[mikesm]

Suse 10.3 alpha6 is out if you want to try that.

Thx
Mike

[KJake]

I'm actually in the process

Having problems with it...can't get any other installation sources to add so I'm stuck with _extremely_ basic packages from the CD at the moment...

[edit]
no editors?? something is seriously wrong with that, vi, vim, joe, nano, pico, emacs, gedit, nothing....
I'm going to wait for the DVD to download via BitTorrent and try installing with that - I was able to get an installation source added, but it wouldn't activate at all, so I couldn't install anything from it.
[/edit]

[mikesm]

Originally Posted by KJake

I'm actually in the process

Having problems with it...can't get any other installation sources to add so I'm stuck with _extremely_ basic packages from the CD at the moment...

[edit]
no editors?? something is seriously wrong with that, vi, vim, joe, nano, pico, emacs, gedit, nothing....
I'm going to wait for the DVD to download via BitTorrent and try installing with that - I was able to get an installation source added, but it wouldn't activate at all, so I couldn't install anything from it.
[/edit]

It's better if you use the DVD... :-)

Thx
Mike

[KJake]

Got the DVD...this is too horribly Alpha for me to feel comfortable running it...plus their kernel doesn't have the PMP patches included.

Thankfully, 2.6.22.1 is considered stable now so there should be an update for the patches soon.

[KJake]

I now have 2.6.22 running with EVMS support, no errors anymore. I just need the PMP patch to get updated and I think I'll be in business (I tried using the pre6 patch but too much changed in pre7 and the final merge).

[mikesm]

Originally Posted by KJake

I now have 2.6.22 running with EVMS support, no errors anymore. I just need the PMP patch to get updated and I think I'll be in business (I tried using the pre6 patch but too much changed in pre7 and the final merge).

Weird. I thought I saw the commit posted for the new PMP patches. Drop Tejun a note and I'm sure he'll respond. You're probably safe building pre6 though... Are you back to Ubuntu?

Did you copy the default config file over when you built a new kernel? I was trying to figure out why you had some many problems with 2.6.18-8 when I didn't have them, and I was wondering if the kernel option settings were messed up somehow.

Thanks,
Mike

[KJake]

It very well could be that I was doing something wrong
I'll try 2.6.18.1 a second and see how it goes.

[edit]
Sure enough, I must have done something wrong before...or it is because I did things differently this time. I started with the oldconfig from a stock 2.6.20 ubuntu kernel and applied the bd-claim patch from EVMS to 2.6.22 (I built 2.6.22 without it first and had errors on the first boot). Then I started menuconfig and configured it for my processor, removed a ton of stuff that I don't need to speed up the building of the kernel (don't need to build all of those modules if you don't need them!). Then, since you suggested 2.6.18.1, I used the oldconfig from 2.6.22 and applied the bd-claim and the libata patches and it built and booted fine.

Here's the kicker :P Looks like my nforce chip isn't compatible with this patch. I have two PMP's, each with 3x500GB attached to them. One goes to the only Sil3132 port I have and the other goes to the SATA1 port of the NVRAID. I'm guessing that my nForce chip is too old. I can't figure out if it has a number, like the 680 I've seen around...it is just labeled as nForce 4 SLI Intel Edition. It's on an Asus P5N32-SLI SE Deluxe.

So, you can see here that 3x500GB is detected on the 3132 and the first 500GB attached to the other one is detected:

root@orion:~# uname -a
Linux orion 2.6.18.1 #1 SMP Sat Jul 21 01:47:09 EDT 2007 x86_64 GNU/Linux
root@orion:~# cat /proc/partitions
major minor  #blocks  name
   8     0  488386584 sda
   8    16  488386584 sdb
   8    32  488386584 sdc
   8    48  488386584 sdd

I'm not really all that upset about this. I have two ports open on the motherboard. I can move those drives all to the motherboard and still make use of all 6x500GB. Looks like I should grab one of the 3132 based controllers from Addonics if I want to use the other port multiplier.

[edit]hmm, not sure that the 2.6.18.1 patch supports AHCI port multiplying actually, that might only be in the latest patches[/edit]

heh, looks like the nVidia doesn't support NCQ...these are the four disks connected to it (the 3 on the 3132 don't error):

Disabling NCQ on sda
/etc/init.d/diskopt: line 47: /sys/block/sda/device/queue_depth: Permission denied
Disabling NCQ on sdb
/etc/init.d/diskopt: line 47: /sys/block/sdb/device/queue_depth: Permission denied
Disabling NCQ on sdc
/etc/init.d/diskopt: line 47: /sys/block/sdc/device/queue_depth: Permission denied
Disabling NCQ on sdd
/etc/init.d/diskopt: line 47: /sys/block/sdd/device/queue_depth: Permission denied

[/edit]

[KJake]

EVMS is all setup and the MD is fully built...so I thought that I'd run IOZone again to test the local disk speed. I'm speechless right now. I'm not sure what's going on, but these are incredible speeds [edit]...near 3Gb/sec? I didn't even think the disks could _actually_ do that yet...[/edit]

This time, I made absolutely sure that I was testing on the MD array, and I bumped the test file up to 4GB max to really thrash the disks. [edit]You can see that after 1GB the speed really starts to drop and at 4GB it drops to ~100MB/sec in Read and ~200MB/sec in Write. This still looks good to me though, should be fine for HD content??[/edit]

It seems that I can't get somewhere without having a problem though...heh. Samba speeds are stuck at 10MB/sec. I've done all the same performance tuning, and it doesn't even seem to make a difference. I tried both NIC ports and get about the same results...I guess I'm going to look over my kernel config and make sure I didn't leave something out...this feels like an overall network layer problem rather than Samba - SFTP seems to be just as slow. If anyone has any ideas, let me know. [edit]I recompiled 2.6.18.1 but without taking out all sorts of things and speeds are still slow over the network. I'm not sure what's going on, but I will be jumping to 2.6.22 when the PMP patch is out.[/edit]

I'll attach the two pretty graphs and for anyone that wants the actual numbers, here's the raw Excel data. [edit]Updated the graphs (again) to look better yet.[/edit]

"Writer report"
        "4"  "8"  "16"  "32"  "64"  "128"  "256"  "512"  "1024"  "2048"  "4096"  "8192"  "16384"
"64"   1033216  1518251  1679761  1722886  1690338
"128"   1291158  1705404  1802755  1905110  1939522  1727352
"256"   1306564  1752173  1855098  1952947  2015259  2034350  1842366
"512"   1323315  1729323  1854787  1947291  1975959  2039772  2014892  1815584
"1024"   1284839  1692394  1805514  1914144  1939207  1995982  1957770  1932227  1651398
"2048"   1245855  1622638  1779631  1854954  1911931  1923059  1887563  1841829  1673209  1285763
"4096"   1083301  1424674  1552658  1611951  1644194  1644351  1627373  1543868  1354900  1135573  870404
"8192"   944888  1156901  1234815  1280416  1302600  1298661  1297827  1263838  1111972  972570  915214  791885
"16384"   894670  1032710  1090806  1133539  1151778  1161492  1166777  1140859  1018280  899825  868725  875075  804086
"32768"   0  0  0  0  1076620  1104310  1114483  1081559  981077  876033  856142  863859  863245
"65536"   0  0  0  0  1067015  1081486  1082342  1065782  961532  855728  836899  848009  850994
"131072"   0  0  0  0  1059501  1069347  1074783  1051251  962420  853716  840955  838999  838667
"262144"   0  0  0  0  942025  732956  909287  926792  833624  739590  761959  784653  719278
"524288"   0  0  0  0  554512  580076  753894  708064  440172  457723  583069  774159  607055
"1048576"   0  0  0  0  533983  549066  572501  398751  549010  462322  456767  477700  440039
"2097152"   0  0  0  0  235181  284124  263830  258056  257516  272716  240105  242036  280956
"4194304"   0  0  0  0  228131  216256  208180  205715  213798  218316  208296  210019  224486

"Reader report"
        "4"  "8"  "16"  "32"  "64"  "128"  "256"  "512"  "1024"  "2048"  "4096"  "8192"  "16384"
"64"   4018152  5860307  5860307  5860307  5860307
"128"   4717434  5545860  5784891  5784891  6727225  6330599
"256"   4477549  5687020  5687020  5938650  6107548  6398720  6249745
"512"   4163357  4827914  5019764  5227492  5495016  5551840  5624545  5439343
"1024"   3955557  4410497  4339202  4898425  5251818  5144870  4949226  4470172  4079544
"2048"   3407614  3828288  3848872  4146110  4248644  4330028  4248644  3828288  3169944  2462141
"4096"   3119752  3497668  3624593  3726809  3838385  3831537  3867766  3720353  2896717  1994225  1579060
"8192"   3091348  3512976  3594563  3666679  3731993  3780445  3934558  3650706  2905979  1943555  1380751  1291728
"16384"   3138763  3578802  3726894  3810380  3881621  3976636  3964248  3853974  3019069  1905338  1339539  1196540  1189313
"32768"   0  0  0  0  4023529  4071082  4118168  3979146  3116048  1939644  1306121  1157879  1161657
"65536"   0  0  0  0  4109045  4191885  4222407  4107080  3233142  1936990  1289623  1131590  1127076
"131072"   0  0  0  0  4221849  4294202  4331384  4248111  3427879  1974039  1293348  1127191  1122074
"262144"   0  0  0  0  4297236  4377005  4403704  4374689  3680462  2027879  1291464  1125129  1118485
"524288"   0  0  0  0  4336649  4426385  4454748  4440104  3814478  2052200  1297120  1126972  1124415
"1048576"   0  0  0  0  4379029  4447562  4483868  4474421  3941616  2080884  1301169  1130640  1124155
"2097152"   0  0  0  0  140054  99301  91799  125835  127368  96230  94511  94158  95288
"4194304"   0  0  0  0  140671  98221  94112  99940  99143  95762  95573  99413  95703

[mikesm]

Originally Posted by KJake

EVMS is all setup and the MD is fully built...so I thought that I'd run IOZone again to test the local disk speed. I'm speechless right now. I'm not sure what's going on, but these are incredible speeds [edit]...near 3Gb/sec? I didn't even think the disks could _actually_ do that yet...[/edit]

This time, I made absolutely sure that I was testing on the MD array, and I bumped the test file up to 4GB max to really thrash the disks. [edit]You can see that after 1GB the speed really starts to drop and at 4GB it drops to ~100MB/sec in Read and ~200MB/sec in Write. This still looks good to me though, should be fine for HD content??[/edit]

It seems that I can't get somewhere without having a problem though...heh. Samba speeds are stuck at 10MB/sec. I've done all the same performance tuning, and it doesn't even seem to make a difference. I tried both NIC ports and get about the same results...I guess I'm going to look over my kernel config and make sure I didn't leave something out...this feels like an overall network layer problem rather than Samba - SFTP seems to be just as slow. If anyone has any ideas, let me know. [edit]I recompiled 2.6.18.1 but without taking out all sorts of things and speeds are still slow over the network. I'm not sure what's going on, but I will be jumping to 2.6.22 when the PMP patch is out.[/edit]

I'll attach the two pretty graphs and for anyone that wants the actual numbers, here's the raw Excel data. [edit]Updated the graphs (again) to look better yet.[/edit]

What distro are you using? Ubuntu or Suse?

What you were seeing with smaller sizes was XFS doing aggressive memory buffering. Still, for an unoptimized array, I think that's pretty good. Run my diskopt.sh script at startup, and I bet you'll see a lot of improvement, though realistically anything over 125 MB/s means you have filled up the gigabit ethernet bandwidth. Still, for copying and other operations, it's nice to have the speed. And why not optimize?

On the Samba issue, have you tested performance with NFS or something else? Did you do the SAMBA optimizations I mentioned in an earlier note?

Are you using the forcedeth driver?

No need to get your 3132 card from addonics, though it certainly works. Any 3232 PCI-E based controller is just as good. This $20 card http://www.newegg.com/Product/Produc...82E16815124027 works just as good. You can get them down to $15 in some places... It's great that 3132 addin cards so SO cheap. And they work great with PMP's.

Thanks,
mike

[KJake]

Originally Posted by mikesm

What distro are you using? Ubuntu or Suse?

What you were seeing with smaller sizes was XFS doing aggressive memory buffering. Still, for an unoptimized array, I think that's pretty good. Run my diskopt.sh script at startup, and I bet you'll see a lot of improvement, though realistically anything over 125 MB/s means you have filled up the gigabit ethernet bandwidth. Still, for copying and other operations, it's nice to have the speed. And why not optimize?

On the Samba issue, have you tested performance with NFS or something else? Did you do the SAMBA optimizations I mentioned in an earlier note?

Are you using the forcedeth driver?

No need to get your 3132 card from addonics, though it certainly works. Any 3232 PCI-E based controller is just as good. This $20 card http://www.newegg.com/Product/Produc...82E16815124027 works just as good. You can get them down to $15 in some places... It's great that 3132 addin cards so SO cheap. And they work great with PMP's.

Thanks,
mike

I'm running ubuntu again, and I've now gone back to 2.6.22-rc6 with the latest libata patch. The server is using the forcedeth driver.

I've recompiled the 2.6.18.1 and 2.6.22-rc6 kernels probably 15 different times now trying to figure out what's going on.

Well...it turns out to be my Vista system again....grr.

I have a brand new Dell system on the Gigabit network in the basement running XP Pro, I decided to finally test it to see if there's something going with my Vista system.

Found a tool called NetIO and ran just basic bandwidth tests.
Vista system on the same Gigabit switch:

TCP connection established.
Packet size  1k bytes:  4499 KByte/s Tx,  7669 KByte/s Rx.
Packet size  2k bytes:  4437 KByte/s Tx,  7766 KByte/s Rx.
Packet size  4k bytes:  4395 KByte/s Tx,  7805 KByte/s Rx.
Packet size  8k bytes:  4351 KByte/s Tx,  8908 KByte/s Rx.
Packet size 16k bytes:  4364 KByte/s Tx,  8905 KByte/s Rx.
Packet size 32k bytes:  4395 KByte/s Tx,  9223 KByte/s Rx.

XP Pro System on a different Gigabit switch:

TCP connection established.
Packet size  1k bytes:  113868 KByte/s Tx,  101442 KByte/s Rx.
Packet size  2k bytes:  114341 KByte/s Tx,  109033 KByte/s Rx.
Packet size  4k bytes:  114368 KByte/s Tx,  112528 KByte/s Rx.
Packet size  8k bytes:  114244 KByte/s Tx,  112157 KByte/s Rx.
Packet size 16k bytes:  114272 KByte/s Tx,  112063 KByte/s Rx.
Packet size 32k bytes:  114332 KByte/s Tx,  112587 KByte/s Rx.

Now THAT is more like it

I just ran another test to my Vista system while I started shutting down different applications (VPN Connection, uTorrent, APCUPSD Tray), and the speed has increased slightly.

TCP connection established.
Packet size  1k bytes:  5574 KByte/s Tx,  9547 KByte/s Rx.
Packet size  2k bytes:  5585 KByte/s Tx,  9792 KByte/s Rx.
Packet size  4k bytes:  5563 KByte/s Tx,  9776 KByte/s Rx.
Packet size  8k bytes:  5630 KByte/s Tx,  11152 KByte/s Rx.
Packet size 16k bytes:  5577 KByte/s Tx,  11136 KByte/s Rx.
Packet size 32k bytes:  5571 KByte/s Tx,  11614 KByte/s Rx.

Man, stupid Vista...at least I can stop compiling kernels looking for an answer to a problem that wasn't actually there. I just don't get what changed on my Vista system to make it go from 50MB/sec to 10MB/sec in a couple of days. I'll keep looking and Googling though, it's gotta be something simple.

Mike, thanks for your help as always.

[KJake]

LOL - figured it out pretty quick...it was iTunes! Yes, really...this is what I've sent to Apple for feedback:

my Music library is on a shared network drive on a Windows XP MCE computer. On my Vista Ultimate system, I map this share to M:. I have iTunes installed on this Vista system and pointed to the iTunes folders and XML library file on this shared drive. Here's the thing, as soon as I play a song, my network connection is limited to 5MByte/sec receive and 10MByte/sec send. It's the oddest thing. As soon as I close iTunes, it jumps back up to ~60MByte/sec receive and ~100MByte/sec send (I'm on a Gigabit network). It doesn't even matter if I pause or stop the song. I also played with looking for shared libraries on my network and sharing my library and turning them on or off didn't make a difference.

I'm using 7.3.1 on Vista Ultimate...pretty odd.

[mikesm]

Originally Posted by KJake

LOL - figured it out pretty quick...it was iTunes! Yes, really...this is what I've sent to Apple for feedback:

my Music library is on a shared network drive on a Windows XP MCE computer. On my Vista Ultimate system, I map this share to M:. I have iTunes installed on this Vista system and pointed to the iTunes folders and XML library file on this shared drive. Here's the thing, as soon as I play a song, my network connection is limited to 5MByte/sec receive and 10MByte/sec send. It's the oddest thing. As soon as I close iTunes, it jumps back up to ~60MByte/sec receive and ~100MByte/sec send (I'm on a Gigabit network). It doesn't even matter if I pause or stop the song. I also played with looking for shared libraries on my network and sharing my library and turning them on or off didn't make a difference.

I'm using 7.3.1 on Vista Ultimate...pretty odd.

Wow... I didnt know vista was so hosed. If I recall, Vista does a bunch of QoS stuff. Look at your vista systems ethernet and tcpip properties and shut off any QoS "value added" functionality. I am thinking that iTunes is requesting dedicated network bandwidth for serving and consuming data. It may be that it's reserving too much or Vista thinks it needs to reserve a lot.

I was going to suggest looking at your Vista client again, but I already have an anti-vista rep on this board. :-)

Thanks
Mike

[KJake]

Originally Posted by mikesm

Wow... I didnt know vista was so hosed. If I recall, Vista does a bunch of QoS stuff. Look at your vista systems ethernet and tcpip properties and shut off any QoS "value added" functionality. I am thinking that iTunes is requesting dedicated network bandwidth for serving and consuming data. It may be that it's reserving too much or Vista thinks it needs to reserve a lot.

I was going to suggest looking at your Vista client again, but I already have an anti-vista rep on this board. :-)

Thanks
Mike

Don't worry, I'm still anti-Vista I just have it because XP would not cooperate on this motherboard and Vista is actually better...but it's networking is really odd. I guess I shouldn't blame it entirely on iTunes since I noticed the same thing happens when I watch a YouTube video, but it is still pretty messed up that I basically can't have iTunes open if I want to move around any big files on the network in a reasonable amount of time.

Another thought is that my network card/driver sucks. I was pretty sure that I have the latest driver, but I'll be checking again. I have a Attansic onBoard Gigabit port on this computer. I learned my lesson and didn't buy another nForce based board...

[edit]Oh yeah, I turned off QoS entirely in my troubleshooting last night (something I do in XP too), but it didn't seem to help. I'm feeling good about the setup now that I've started to migrate all the files on my existing 2TB array to it so that I can move the 2TB array disks into this new server too. Then I'll have one nice, big network share where pretty much everything will be stored. I'll need to come up with some sort of backup solution soon...I didn't have one for the 2TB array, but used Volume Shadow Copies in Vista and that saved my butt a couple of times when I accidentally deleted something. We could have this function in Linux soon http://ubuntuforums.org/showthread.php?p=2850968.[/edit]

[mikesm]

Originally Posted by KJake

Don't worry, I'm still anti-Vista I just have it because XP would not cooperate on this motherboard and Vista is actually better...but it's networking is really odd. I guess I shouldn't blame it entirely on iTunes since I noticed the same thing happens when I watch a YouTube video, but it is still pretty messed up that I basically can't have iTunes open if I want to move around any big files on the network in a reasonable amount of time.

Another thought is that my network card/driver sucks. I was pretty sure that I have the latest driver, but I'll be checking again. I have a Attansic onBoard Gigabit port on this computer. I learned my lesson and didn't buy another nForce based board...

[edit]Oh yeah, I turned off QoS entirely in my troubleshooting last night (something I do in XP too), but it didn't seem to help. I'm feeling good about the setup now that I've started to migrate all the files on my existing 2TB array to it so that I can move the 2TB array disks into this new server too. Then I'll have one nice, big network share where pretty much everything will be stored. I'll need to come up with some sort of backup solution soon...I didn't have one for the 2TB array, but used Volume Shadow Copies in Vista and that saved my butt a couple of times when I accidentally deleted something. We could have this function in Linux soon http://ubuntuforums.org/showthread.php?p=2850968.[/edit]

Sounds like you've made a lot of progress. I guess we should be too mad at PC vendors for not making gear that pushes things around at a gigabit, who would have thought that a few years ago, consumers would have gear that could push things that fast?

Also, note that EVMS has snapshotting capability. Makes it easy to run backups of live filesystems...

Thanks
Mike

[KJake]

Originally Posted by mikesm

Sounds like you've made a lot of progress. I guess we should be too mad at PC vendors for not making gear that pushes things around at a gigabit, who would have thought that a few years ago, consumers would have gear that could push things that fast?

Also, note that EVMS has snapshotting capability. Makes it easy to run backups of live filesystems...

Thanks
Mike

Cool, I'll have to read up on that. I have my 4x750gb drives clean and installed in the case, but nothing to connect them to. I realized that when I first tried the PMP on the NV ports was when I was running 2.6.18.1 and I wasn't sure that AHCI PMPing was supported with that patch...so I had a good feeling about the 2.6.22-rc6 since the release notes are explicit in what it now supports. Alas, no such luck. I'll have to grab a controller (perferably looking for a 4 port SATAII).

I put those 2x250gb in my desktop setup for RAID1 so that I still have a decent amount of storage on my desktop, for faster disk access and some data guarding at the same time.

[mikesm]

Originally Posted by KJake

Cool, I'll have to read up on that. I have my 4x750gb drives clean and installed in the case, but nothing to connect them to. I realized that when I first tried the PMP on the NV ports was when I was running 2.6.18.1 and I wasn't sure that AHCI PMPing was supported with that patch...so I had a good feeling about the 2.6.22-rc6 since the release notes are explicit in what it now supports. Alas, no such luck. I'll have to grab a controller (perferably looking for a 4 port SATAII).

I put those 2x250gb in my desktop setup for RAID1 so that I still have a decent amount of storage on my desktop, for faster disk access and some data guarding at the same time.

PMP's are only supported on SI3124/3132 ports and AHCI ports. I am pretty sure the NF4 SATA ports aren't AHCI. If the kernel probes them as sata_nv they aren't AHCI.

I don't think you can find a 4 port SATA II PCIE adapter with the 3132 (it only supports 2 ports). I'd get a couple cheap syba (or the like) 3132 adapters. You can get them for $15-$20 a piece.

BTW, you can plug the disks into the sata_nv ports and they will work just fine, even if some of the other disks in the raid array are on PMP ports. No issues with mixing and matching under linux. If you add a 3132 later, you can simply unplug them from the nv ports and plug them into the PMP prts hung off the 3132. The kernel will read their id's and reassmble the disk array on the new ports automatically. Try that with hardware raid. I know it works as I've done it myself a couple times.

thanks,
Mike

[KJake]

Originally Posted by mikesm

PMP's are only supported on SI3124/3132 ports and AHCI ports. I am pretty sure the NF4 SATA ports aren't AHCI. If the kernel probes them as sata_nv they aren't AHCI.

I don't think you can find a 4 port SATA II PCIE adapter with the 3132 (it only supports 2 ports). I'd get a couple cheap syba (or the like) 3132 adapters. You can get them for $15-$20 a piece.

BTW, you can plug the disks into the sata_nv ports and they will work just fine, even if some of the other disks in the raid array are on PMP ports. No issues with mixing and matching under linux. If you add a 3132 later, you can simply unplug them from the nv ports and plug them into the PMP prts hung off the 3132. The kernel will read their id's and reassmble the disk array on the new ports automatically. Try that with hardware raid. I know it works as I've done it myself a couple times.

thanks,
Mike

Ha, yeah, I tried that with hardware raid once...boy was I mad for a while after I lost all the data. I took it pretty well considering my primary hard drive had just died recently too.

I wasn't aware of the 2 port limitation of the 3132...but that makes sense why on my board there are only 2 ports supported for the Sil RAID (one port is eSATA so I can't really make use of it - except for an external set of disks).

I've got 2 x1 PCIe slots that aren't used, so I can get my 4 ports that way. There's an x8 slot too that would work if I ever needed more, but I'm not sure that I could actually fit that many disks in here (if I attached a PMP to each native port).

Thanks for the Syba link, I'll probably order one to start since that's all I need for now.

[BobPhoenix]

Originally Posted by mikesm

The kernel will read their id's and reassmble the disk array on the new ports automatically. Try that with hardware raid. I know it works as I've done it myself a couple times.

I just got done doing this with my two 3Ware cards. Moved drives from one computer to another on different 9500S cards (Windows2000 Server to WindowsXP as well). And the unit came up just fine. In fact I had to destroy it because I wasn't wanting to transfer the data just the drives. So while some hardware raid might have problems I bet all the top brands like Areca/3Ware can do it.

[KJake]

Originally Posted by BobPhoenix

I just got done doing this with my two 3Ware cards. Moved drives from one computer to another on different 9500S cards (Windows2000 Server to WindowsXP as well). And the unit came up just fine. In fact I had to destroy it because I wasn't wanting to transfer the data just the drives. So while some hardware raid might have problems I bet all the top brands like Areca/3Ware can do it.

It typically does work just fine when you use the same hardware. Mike's point was that I would be moving the drives to different controllers (nVidia to Sillion Image) and that if you try to do that with hardware raid controllers it won't work...which I confirmed in my reply. I moved drives from an NV raid to a HighPoint and the data was history.

[mikesm]

Originally Posted by KJake

It typically does work just fine when you use the same hardware. Mike's point was that I would be moving the drives to different controllers (nVidia to Sillion Image) and that if you try to do that with hardware raid controllers it won't work...which I confirmed in my reply. I moved drives from an NV raid to a HighPoint and the data was history.

Exactly. Try moving your data disks from a 3ware controller to an Areca and keep all the data. That's the issue.

Thanks,
mike