[Phoenix32]

Originally Posted by blue68f100

Phoenix32 may correct me on the 4000's issues.

Hmmmm, what can I add? Okay, here goes...

Much like the 4100, there is some stupid crap going on inside the 4000 that makes no sense. There is no binary reason I can figure out why there should be a limit at 1 TB (or 1.2TB) with the 4000. It would make a lot more sense if it was at around 2.1 TB (I think it was 2.1, I just don't feel like doing the math right now).

Now with that said, you yourself are reporting errors with 4 x 300 GB drives? Why? Can you tell me? Probably not. There is a good chance it has nothing to do with the size itself, but it could be. Why? Because there are some who say 4 x 300 works fine with no errors. There are some who say 4 x 320 doesn't work at all and yet others who say they have 4 x 320 with no errors or problems. To the OS, a RAID 0 or a RAID 5 should make no difference (concerning size) and the SNAP 2000 (some of them, but not all) are based on the same type board and a 2000 user tried 2 x 750 (1.5 TB) and it gave him errors saying it was a proposterous number. So you tell me. So is it a power supply problem? Is it an OS problem? Is it a hardware limitation? Is it just a bone head user making assumptions? Is it certain brands of drives? Certain revisions? Etc Etc Etc... Too many questions and not enough answers...

Here is what I can tell you.

Looking at the different revisions (and I have seen all 4 of them), there seems to be no real difference in the chips or chipset being used or anything that should affect one doing a size larger than another revision. Please note I said should, but there is some evidence to back it up.

Two of the revisions use M/S (master slave) and two of them use CS (cable select). There is a stupid issue with the 4000 and it's hardware that when a drive in a M/S setup, depending on how it dies (if it is electronics, not physical head or disk) or if you power the unit up without it's partner, will drop the partner drive out as an orphan also. We all know what happens when you orphan 2 drives (the bad one and it's partner) in a RAID 5 array. If you don't, then ignore the rest of this message. The cure is to use CS, but this brings in a new problem. Sometimes, even with the proper cables etc, if you put the drives in CS, you will get random drive errors and dropping drives in and out of the array (which is useless). Now the interesting part of this is that it never does this with the set of original quantum 30 GB drives I have. This causes me to believe that it has to do with the drives themselves, not the 4000. Why? Because as a tech, I know for fact that some drives, when in CS mode, will not automaticly drop their interface speed down the current PATA/IDE interface speed being used. Meaning an ATA100 drive dropping to the ATA33 or ATA66 your controller might be using. I strongly suspect that is what happens here. You put the drives in CS, and with all the newer drives being ATA100 or ATA133, some of them do not downshift to the 4000 controller speed, so you get random errors. Can I prove it? No, but it is a damn good bet this is the real problem. Either way, on some 4000's, you are stuck with M/S, with some drive setups.

Then there is the beloved power supply. Now nobody is going to want to hear this, but here it is flat out. From a hardware spec point of view, you cannot use 4 x 250 GB drives (or similar) in the 4000. Why? Because the power supply is just not rated for it, that's why. The SNAP 4000 power supply is rated at 6A on the 12V rail. Look at the specs of any of the larger drives and you will fast see that when they are running, you are running the power supply at anywhere from 75% to 95% of rated spec 24/7. That is NOT recommended in any power supplies. A max rated spec is not for continuous use. But skip that for a moment. The big issue is spin up power. Look at the spin up current required for most modern larger drives. The Seagate 250 GB drive is 2.8A for an example (A very common drive used by 4000 owners). 4 x 2.8 = 11.2A. Ummm last time I checked, 11.2 is significantly larger than 6 (almost double). Now here is the up side to this. The power supplies used in the 4000 seem to be pretty solid electronics and let us get away with it for these short bursts and long term usage., but just KNOW that you are far exceeding rated spec.

Keeping on the power supply, we also know these 4000 units are getting long in the tooth (old). Meaning these power supplies get weaker over time and use. If you get a 4000 with a marginal power supply, what will happen is you will see random drives drop in and out when you power the unit up. It will drive you nuts trying to figure out what is going on, and all along it is a marginal power supply causing the issues. Trust me, I have seen this in 2 different units first hand and it sucks.

Now I know this is about the 4000, but keep in mind what I am about to say for the next paragraph(s). In the 4100, we have seen people use 160 GB drives quite often and get the max 137 GB of the unit. But every once in a while, a user will tell us his won't work. Every time I can think of at the moment, it has been with maxtor drives. Interesting isn't it (in a Spock kinda way)? Sounds to me like the problem is not the user's 4100, but rather maxtor, or at least certain revisions of maxtor 160 Gb drives. Why? Because when a few of these users have swapped to other brand 160 GB drives, it all worked fine.

Now we get to the REAL PROBLEM. The real problem is the people reporting their results. People do not test and report properly. Some guy sticks 4 hard drives in CS mode into his 4000 and it get's all flakey, so he reports "You can't use drives in CS mode in the 4000". BULLCRAP! 2 of the revisions of the 4000 came with drives in CS and the other two revisions will in fact use cable select with the right cable and drives. The problem was his conclusions. His problem may lay elsewhere, like the drives maybe, and their ATA speed, or the cable, hint hint. Then another guy sticks 4 x 250 GB drives in his 4000 and it goes all flakey, so he reports, "You can't use 4 x 250 GB drives in a 4000". In both cases, they reported it as fact. Well guess what? Plenty of people use 4 x 250 GB drives in their SNAP 4000. Maybe the problem is your power supply being marginal with those drives or the model of hard drive? But, but, but, but, my 4 x 80 GB drives worked fine and these 250's, same brand, don't. Well #1, same brand does not mean they work exactly the same. Hard disks, even of the same model number sometimes, go through plenty of revisions and these revisions change things. Then #2, even if they are in the same family, and rated the same for specs, I can promise you, that those 80 GB drives draw a lot less power during spin up than those 250 GB drives do. It's called physics (mass and energy). Then there is the opposite. Someone does something, and it appears to work, so he comes back and says "such and such works in the 4000", like it is fact. Then later, when one of the drives fails or something, he replaces the drive and the array won't rebuild. Maybe because of something else, but also maybe because his conclusions about such and such working may have been wrong. This is the case with the 4 x 300 Gb and 4 x 320 GB drives and why David and I refuse to say it does or doesn't work. Just not enough testing completed to prove it out.

So the REAL PROBLEM is people reporting their results. They see one result and then ASSUME things and report that thing as FACT. That is why there has been so much confusion on this forum about a great many subjects. Too much assuming going on and not enough COMPLETE testing. THEN, when you ask someone to try this or that or to test something, quite often, the reply is, "sorry, I don't have the time" (after you wasted your time helping them) or they never come back and report what happened. They get what THEY NEED and then disappear. Sorry, truth is truth and I am 100% sure the regulars around here will back me up on that. The long and short is that most of us just do not have the money to go out and buy every SNAP type and revision as well as all kind of disk sizes (in sets) to test and figure out where limits are and what does and does not work. For example, David and I have a working theory on the 4 x 300, 4 x 320 GB, and 4 x 400 GB drive setups. But guess what? David does not have a 4000 and I cannot afford the drives to do the testing myself on my various 4000 units. So unless we can find someone who has these things and is willing to test TOP TO BOTTOM for facts, we just have to continue to give GENERAL parameters and elusive answers. We are doing our best.

NOW, before someone says, "There goes Phoenix on a rant again". I am not ranting or even a little bit ruffled right now, honest. I am just answering the question I was directly asked by name and being totaly open and honest about it. Admin69, please do not think I am ranting at you. I am just trying to clue you in to how elusive some of these answers you want really are, and why.

What I can tell you is; (and yes, it changes over time with real results aquired and new theories)

- 4 x 250 GB works in the 4000
- 4 x 300 and 4 x 320 should work in the 4000
- 4 x 400 and even 4 x 500 might work in the 4000
- 4 x 750 will probably not work

- the 4000 power supply is over spec with modern large drive sizes
- the 4000 will act flakey and random if the power supply is marginal (and this includes maybe even sending errors on startup)

- M/S setups in the 4000 are very risky with RAID 5 setups if a drive fails
- CS setups in the 4000 work just fine, but may require a cable modification and may have problems with some drives (brands or models)

- I seriouly doubt we will every know what the real story is on some of the "whys" because it is buried deep in the OS, BIOS, and built in drivers which we do not currently have access to nor probably ever will.

- Guardian units are way to noisy (sound like hair dryers) for most home users use (this one was just for David)


There ya go....