Help with 4000 upgrade

[Phoenix32]

Originally Posted by blue68f100

From what users have reported, the 300 formated fine, when they went to build the array (4 disk) is when all the problems showed up. But again he is using seagates, which has have the fewest problems. Those are the new 7200.10 Perpendicular drives.

What version of OS are you using, as well as HW, I think all of the bios are the same.

Hmmmmmm ???

I think that was what he was saying Dave, that he was configuring the 4 disk RAID 5 and it was 27% done in 30 minutes. Maybe not???

I gotta disagree on the Seagate thing though, sorry Dave. What brand hard drive you are using should make no difference at all in this other than the normal speed differences between the those brand hard disks. Sorry, I just don't buy it.

This last part is where I would suspect the difference would be. The circuit board revision, the BIOS revision, and the OS revision. This right here is why it totaly sucks that Adaptec is so poor about providing any real information about the SNAP. It would be nice to have a revision history for the HW/BIOS/OS including what changes were made. For Example; as you know, I mentioned once before that I suspected there was more difference between 3.4.805 and 4.0.860 than just MS AD. You just have to wonder about things like support for over 1.0 or 1.2 TB and little things like that, but we will never know for sure without a revision history for these things. When it doesn't work, we need to look at the HW/BIOS/OS revision, and when it does work, do the same (which is what I think you do already Dave).

[pocketrocket]

Well it finished without a problem. Looking at the log the process started at 11:59am and finished at 2:32pm that is 2 hrs 33 min. Not bad at all.
I now have raid 5 with 898,399 total disk space. WOW this is great I am now going to upgrade our last snap 4000.

_________________________
1 Snap 4000 - 80gig (4 x 30gig quantum's) v4.0.860 Raid 5
1 Snap 4000 - 700gig (4 x 250gig Seagate's) v4.0.860 Raid 5
1 Snap 4000 - 900gig (4 x 320gig Seagate's) v4.0.860 Raid 5

[blue68f100]

That's good news, would be interesing to see how the new perpindicular drive workout. I need to dig up the head count, cyl, platters and compare to the older models. I now the density went way up with this new technology. I think the 300 is done on 2 platters. I also know that users have had problems with the 300 gig drives. Do not know if was related to their hardware, bios, HW revisions.

Are your 250 gigs the 7200.9 drives?

Pheonix, I'm still going to disagree with you. I still believe it is a hardware hd, cyl, platters that are cause the problem. I think the OS has limits with the values. It may all be BIOS related. I do know that most of the unix flavors now recomend setting the bios to NO DRIVES. This way the OS reads the drive data, not BIOS inturpeted. But since we don't have access to SnapAppliance engineering data, We may Never NO for Sure.

Unless one of their engineers that were layed off due to adaptec purchase finds this forum and en-lightens us. What are those odds????

Edit: Pulled Seagates specs on the 320gig 7200.10 drives
4 heads
2 platters
2.8 amp Startup Current on 12vdc

It use to take 4 platters to get this capacity 2 yrs ago. 3 platters 1 yr ago, How long to 1 platter ???

[pocketrocket]

seagate 7200.8 ST3250823A 250gig 2 platers
7200.10 ST3320620A 320 gig Perpendicular Recording 2 platers

[blue68f100]

Looks like the 300 gig models of the 7200.8 use 3 platters.

[Phoenix32]

Originally Posted by blue68f100

Pheonix, I'm still going to disagree with you. I still believe it is a hardware hd, cyl, platters that are cause the problem. I think the OS has limits with the values. It may all be BIOS related. I do know that most of the unix flavors now recomend setting the bios to NO DRIVES. This way the OS reads the drive data, not BIOS inturpeted. But since we don't have access to SnapAppliance engineering data, We may Never NO for Sure.

I understand, but "most" hard drives these days are pretty close to the same number of cyl/hd/platter combinations from each mfg. I do agree that we will probably never know "exactly", but we can hope, or at least narrow it down as much as we can...

[Phoenix32]

Originally Posted by pocketrocket

seagate 7200.8 ST3250823A 250gig 2 platers
7200.10 ST3320620A 320 gig Perpendicular Recording 2 platers

Can I assume the "7200.10 ST3320620A 320 gig Perpendicular Recording 2 platers" were what the 4 drives in the 4000 were that worked?

And just in case I missed it, What HW/BIOS/OS/Model you using (by model I mean -001 -002 -003 or -004 revision)?

I ask because I want to compare it to my 4000 so I have a good idea of what I can get away with putting in there...

[pocketrocket]

1 Snap 4000 - 80gig (4 x 30gig quantum's) v4.0.860 Raid 5
MDL 70700069-001 HW 2.0.1 Bios 2.0.282

1 Snap 4000 - 700gig (4 x 250gig Seagate's) v4.0.860 Raid 5
MDL 70700042-001 HW 2.0.1 Bios 2.0.252

1 Snap 4000 - 900gig (4 x 320gig Seagate's) v4.0.860 Raid 5
MDL 70700042-002 HW 2.0.2 Bios 2.0.252

Does anyone know how to check for the amount of ram installed using the debug?

[blue68f100]

"debug memory"

[pocketrocket]

Thanks!
How did I miss that? Is was to easy.

[Phoenix32]

Originally Posted by pocketrocket

1 Snap 4000 - 900gig (4 x 320gig Seagate's) v4.0.860 Raid 5
MDL 70700042-002 HW 2.0.2 Bios 2.0.252

Hmmm, mine is

MDL 70700042-004 HW 2.0.1 BIOS 2.0.282 and I am using v3.4.805 OS.

Whataya think David? My model is a newer revision, but and older HW version number, with a newer BIOS and he is using v4 OS while I am using v3. Think I would get the same results with those Seagate 320 drives?

BTW pocketrocket, you did not answer my other question. These Seagate 320 drives you are using. They are the "7200.10 ST3320620A 320 gig Perpendicular Recording 2 platers" correct? (I hate making assumptions, always costs me money somehow)

[pocketrocket]

Originally Posted by Phoenix32

Hmmm, mine is

MDL 70700042-004 HW 2.0.1 BIOS 2.0.282 and I am using v3.4.805 OS.

Whataya think David? My model is a newer revision, but and older HW version number, with a newer BIOS and he is using v4 OS while I am using v3. Think I would get the same results with those Seagate 320 drives?

BTW pocketrocket, you did not answer my other question. These Seagate 320 drives you are using. They are the "7200.10 ST3320620A 320 gig Perpendicular Recording 2 platers" correct? (I hate making assumptions, always costs me money somehow)

Phoenix32
Yes this is the drive i have used. So far it works just fine. I have rebooted the snap a few times and it always restarts. I think you need to upgrade your OS.

[Phoenix32]

Originally Posted by pocketrocket

Phoenix32
Yes this is the drive i have used. So far it works just fine. I have rebooted the snap a few times and it always restarts. I think you need to upgrade your OS.

I have 4.0.860, however, I have not installed it because the recommendations around here have been that it is a waste of time if I am not going to use MS AD. I originaly was going to be using MS AD, but things change, now I won't be, so....

Why do you think I need to up the OS to v4? Never hurts to ask...

[pocketrocket]

Well I think the first reason is because it works.
and I have it on 2 machines with large drive with 100% success.

[radio]

Originally Posted by blue68f100

Well if you ran 2 mirrors. You would have about the same capacity. Would be slower comparied to raid 5. But on a disk failure you can still read the data.

I am not sure I understand this. I had always assumed that you could still use a RAID 5 before and while rebuilding after one of the drives failed. correct?

[blue68f100]

Yes, You will operate in degradded mode till the drive is replaced and resynced in to the set. The problem with a raid 5 is, if you loose another drive before you get it replaced your SOL. And Degrade speed is 25% ?? of normal.

It's not any different than running a mirror (raid1) execpt there is no degrade mode. Works are normal speed. I think the 4x00 can run 2 raid 1 shares. You loose some speed, but may be able to load them with 4x 750gig drives. Thats like 2 x 720gig usable space. Just if it had to power to spin them up.

ALWAYS setup the email notification on any network device, and test to make sure it works. The earlier you know you have a problem, the better off you will be.

I thought I read some where that the best raid 5 performace accours with 7 drives, then have a hot spare. I don't like a hot spare if its spinning at speed. It should be in hibernation mode till needed.

[radio]

Originally Posted by blue68f100

You loose some speed, but may be able to load them with 4x 750gig drives. Thats like 2 x 720gig usable space. Just if it had to power to spin them up.

Yeah, I guess I would like to know that someone has tried a 750GB drive in a snap 4000 before spending $1300. Besides, I could just get two $30 promise cards to mirror them rather than the snap.

If I loose two drives there, I have a 33% chance of being 50% SOL or a 66% chance of getting lucky.

[Phoenix32]

Originally Posted by blue68f100

You loose some speed, but may be able to load them with 4x 750gig drives. Thats like 2 x 720gig usable space. Just if it had to power to spin them up.

Ahhhh, but THAT is the question of the day (other than the magic number). With the new perpendicular drives using less platters, and each company having different power-up and normal use power requirements, we are back to the, "how much can we put in one without changing the power suppy" question. I don't think we ever did decide if it was 2 or 3 platters you could get away with in a 4000 (not having to do with size, but rather spin-up power). If we did, I missed it somehow. But I do recall we decided 4 platter drives was too much and didn't work, I think. Didn't we? So is it 2, or is it 3 platters as the limit? Then, based on that, how much space can we get within those platters limits with the newer perpendicular drives?

Do not hold me to this, but I am guessing it is going to be as follows.

for RAID 5, 320GB drives is going to be the limit (1280GB or 1.28TB which makes sense math wise).

for RAID 0, same limits as RAID 5 and for same reasons.

For RAID 1 and JBOD or combinations, it will be 3 platters (as long as they are quality efficient drives) which I think right now translates to about 400GB drives but will soon be 500GB drives.


As an off the wall option, you could go to 4 x 400GB drives, with only 3 hooked up, using the 4th installed, but not connected, as a MANUAL SPARE. Then set the 3 drives as RAID 5 for about 800GB or so and if/when a drive fails, you have a brand new never used spare sitting right there that all you have to do is hook up and disconnect the bad drive. This SHOULD stay within the magic number barrier as well as stay under the spin-up power limits.


What-a-ya think of my ramblings there Dave (you too Jontz so you don't feel left out)?

[blue68f100]

I believe you are in line with what I was hinting. It does get back to the PS & the magic number. Pulled the spec sheet for the 7200.10
320gig 2 platters
400gig 3 platters
500gig 3 platters
750gig 4 platters

The .8 and .10 pull the same startup current on the 12 vdc 2.8A, I the difference is how long it stays at these high loads.

[jontz]

Originally Posted by Phoenix32

What-a-ya think of my ramblings there Dave (you too Jontz so you don't feel left out)?

My head hurts...

Actually, I agree with you. I like the "manual spare" theory, althought I am starting to question the logic of using a snap 4000 if you need more than 1.2 TB of space...it seems to me that you would be better off using a normal PC with a gigabit NIC card in it running linux at that point...

but that's not why we are here, is it?

[blue68f100]

If you look around at all of the RAIDs out there, most do not break the 1T limit unless it done with a expansion unit. Then you are talking about big bucks.

This is the sole reason I'm playing with the FreeNAS option. Drives are getting cheaper. Using a 8 or 16 SATA controller, for expansion as needed. But you need to decide what drives and buy a case. This way everyone will be compatable with each other.

The bigger problem is that you still need to backup the data somehow.

[Phoenix32]

Originally Posted by blue68f100

If you look around at all of the RAIDs out there, most do not break the 1T limit unless it done with a expansion unit. Then you are talking about big bucks.

This is the sole reason I'm playing with the FreeNAS option. Drives are getting cheaper. Using a 8 or 16 SATA controller, for expansion as needed. But you need to decide what drives and buy a case. This way everyone will be compatable with each other.

The bigger problem is that you still need to backup the data somehow.

You meant 1.2Tb right?

As for FreeNAS, have they solved the RAID 5 problem yet? Last I checked, about a little over a month ago, they found a problem where RAID 5 works, but after a bad drive replacement, it won't rebuild the RAID, completely defeating the whole purpose in using RAID 5. This is why I did not go FreeNAS in fact.

[blue68f100]

Actual usable space at 1T.

They add another raid system to the one they were working on. The one added was one that had been used with most Linux file systems. I think the XFS had a patton that expired, the reason we are starting to see it in the free flavors of unix. The one they were having a problem with was the XFS format that was just added to FreeBSD. FreeBSD 6.0 only could read the XFS. They were trying all new stuff. It only seam to be a problem with the software raid. Hardware was working fine. Mine is still setup with the XFS structure. When go to the drive and RAID setup now, you have more options than carter has pills. But only one is consider stable.

[blue68f100]

Originally Posted by jontz

My head hurts...

Actually, I agree with you. I like the "manual spare" theory, althought I am starting to question the logic of using a snap 4000 if you need more than 1.2 TB of space...it seems to me that you would be better off using a normal PC with a gigabit NIC card in it running linux at that point...

but that's not why we are here, is it?

I would expect the drive capacity to start ramping up and past the 1T capacity in 6-12 months. So you can buy any MB that has SATA II Raid controllers with dual gigabit networks. Drop 4 1.2T drives in raid 5. You have a 4 T NAS with dual gig speed. With FreeNAS the frontend. Total setup time < 1 hr.

Then we want need our snaps.

[Phoenix32]

Originally Posted by blue68f100

Actual usable space at 1T.

Actual usable space I think is even less than that... Like about 900MB, close enough...