Pro/Forums - View Single Post

Phoenix32 · 03-02-2007, 04:57 AM

Why thank you David, you are too kind...

1. Yes, the 4000 will run 4 x 250 GB drives no problem.

2. No, it is not linux, it is a bastardized version of BSD, which itself is a bastardized version of unix.

3. LBA48Bit has to do with the addressing. Prior to LBA48bit, 28 bits was used instead, which limted hard disk sizes to < 137 GB (at least without some trickery going on). So in other words, a system needs to support LBA48bit to be able to address drives larger than 137 GB. Now support means the OS has to support it and so does/do the drives controller(s). In the case of the SNAP 4000, it's controllers do support LBA48Bit and SNAP OS revisions 3.4.805 and above also support it.

As for the revision required, well, any of the SNAP 4000 can do 4 x 250 GB, not just -003 and -004. When we say revision, we are talking about the -00x number on the end of the model number (-001 -002 -003 and -004). So don't wory about that to0 much.

What David was talking about is a little bug for RAID 5 (if you want to call it that) in the drive setup for revisions -001 and -002. These two revisions used a Master Slave setup while the -003 and -004 revisions used C/S (Cable Select) type setup. What difference does that make you ask? Well I will tell you here. (if your not interested, skip down 3 paragraphs)

In a Master Slave setup, if a drive fails, it might, not will, cause you to loose the whole RAID 5 array (which means all your data). There are two ways this can happen. 1) If you remove the bad drive and then power the unit on without installing a new drive (properly). 2) If the drive fails in the electronics in such a manner as to make the drive invisible to the controller (remember, most drive failures are disk and head related, but not all). If the drive does not disappear to the controller, -AND- you replace the drive properly without powering the unit up while the bad drive is remove, there should be no problem. What happens here is that when oen drive disappears or is removed (same thing), the controller gets confused and takes the other (good) drive on that controller and orphans it also. 2 Orphans in a 4 drive RAID 5 = bye bye RAID 5 and data.

In revisions -003 and -004, C/S is used and if a drive goes away on a controller, it still sees the other drive on that controler just fine, thus no problem. So the solution (work around) is to use C/S. But there is a catch to that too. For reasons I am not going to go into here (it is way too in depth), I will just tell you this. Most Modern drives that are ATA 100 and ATA 133 (that's about all of them these days), when put in C/S with a standard cable will have issues down shifting to the speed of the controller as they do so nicely in a Master Slave setup. So witht he older controllers in these SNAP 4000 units, guess what happens? Yup, you got it, things puke and drives disappear or are not seen at all (weird crap happens).

Now there were some key words there above. "With standard cables". The -003 and -004 for the most part use the same controllers and quite often even the same revision main boards, so how do they get away with doing C/S then when the -001 and -002 won't (with modern drives)? Damn good question, and I got your answer. They don't use a standard cable. If you look close at the cable on a -003 or -004 unit, you will notice that there is a cable lead cut going to the second drive on the chain. This little removal of that one elad makes all the difference in the world to the controller and hard disk(s), again, for reasons I won't go into here. So, the FIX for a -001 or -002 revision SNAP 4000 so it won't have this little bug/problem is to use C/S and to remove that one little lead on the ATA cables. That simple. And trust me, it is not hard at all.

So bottom line is that you can use 4 x 250 GB drives, and do it with any revision SNAP 4000, so long as you do the proper things needed. Like update the OS if needed and change the drive configuration if it is a revision -001 or -002.

Okay, David also mentioned the Power Supply. While I have been chastized for saying this before, it is still true and I will repeat it here. The SNAP 4000 power supply is just not designed for 4 modern larger drives. It is not so much the issue of running the drives, it is the issue of spinning them up on initial power up (the first couple seconds). According to spec, a Seagate 250 GB drive needs 12V @ 2.8A during spin up. Multiply that times 4 and you have 11.2A. Last time I checked, the SNAP 4000 power supply is rated at 6A on the 12V rail. You do the math.

Now, in fact, the SNAP 4000 does not have a weak power supply, it has a damn good power supply. Why? Becasue even under those peak loads, a SNAP 4000 power supply will usually handle the momentary load and do the job. BUT, and you knew there had to be a but, these SNAP 4000 units are getting older by the day, so it is not uncommon to have a power supply that just can't handle that anymore. It happens with age on power supplies. So be prepared for it.

By the way, replacement power supplies for the SNAP 4000 are hard to come by and are expensive when you do find them. I have come up with a MOD where you can use an inexpensive AT or ATX power supply as a replacement, but you need to know how to solder, understand wiring diagrams etc, and have a cutting tool like a dremel. It is not for the just anyone to do. Northwest Technical does have a replacement power supply they sell for $150 at http://www.northwesttechnical.com/

If you use larger drives, you will most likely need to upgrade the memory to 128 MB if it's not already, especially if you load JVM. If you don't, when you try to tansfer large amounts of continuous flow data (large files), you will get errors and the transfers can/will fail.

There, now you should be armed with the knowledge you need to make a 1000 GB (1 TB) SNAP 4000.

If you have questions or need help, just ask... I have 6 SNAP 4000 here and have done a ton of testing on them. Chances are, if it's hardare on a 4000 (and even on the 4100), I have seen it already. And if I don't know, I am sure David, Sam, Jontz, Hallis, or one of the other many good people here will. Note: I am a Hardware guy, not an IT, networking, or security guy.

Hey David, can you tell I took my happy pills today?

03-02-2007, 04:57 AM	#3
Phoenix32 Thermophile Join Date: May 2006 Location: Yakima, WA Posts: 1,282	Re: Question about the Snap 4000 Why thank you David, you are too kind... 1. Yes, the 4000 will run 4 x 250 GB drives no problem. 2. No, it is not linux, it is a bastardized version of BSD, which itself is a bastardized version of unix. 3. LBA48Bit has to do with the addressing. Prior to LBA48bit, 28 bits was used instead, which limted hard disk sizes to < 137 GB (at least without some trickery going on). So in other words, a system needs to support LBA48bit to be able to address drives larger than 137 GB. Now support means the OS has to support it and so does/do the drives controller(s). In the case of the SNAP 4000, it's controllers do support LBA48Bit and SNAP OS revisions 3.4.805 and above also support it. As for the revision required, well, any of the SNAP 4000 can do 4 x 250 GB, not just -003 and -004. When we say revision, we are talking about the -00x number on the end of the model number (-001 -002 -003 and -004). So don't wory about that to0 much. What David was talking about is a little bug for RAID 5 (if you want to call it that) in the drive setup for revisions -001 and -002. These two revisions used a Master Slave setup while the -003 and -004 revisions used C/S (Cable Select) type setup. What difference does that make you ask? Well I will tell you here. (if your not interested, skip down 3 paragraphs) In a Master Slave setup, if a drive fails, it might, not will, cause you to loose the whole RAID 5 array (which means all your data). There are two ways this can happen. 1) If you remove the bad drive and then power the unit on without installing a new drive (properly). 2) If the drive fails in the electronics in such a manner as to make the drive invisible to the controller (remember, most drive failures are disk and head related, but not all). If the drive does not disappear to the controller, -AND- you replace the drive properly without powering the unit up while the bad drive is remove, there should be no problem. What happens here is that when oen drive disappears or is removed (same thing), the controller gets confused and takes the other (good) drive on that controller and orphans it also. 2 Orphans in a 4 drive RAID 5 = bye bye RAID 5 and data. In revisions -003 and -004, C/S is used and if a drive goes away on a controller, it still sees the other drive on that controler just fine, thus no problem. So the solution (work around) is to use C/S. But there is a catch to that too. For reasons I am not going to go into here (it is way too in depth), I will just tell you this. Most Modern drives that are ATA 100 and ATA 133 (that's about all of them these days), when put in C/S with a standard cable will have issues down shifting to the speed of the controller as they do so nicely in a Master Slave setup. So witht he older controllers in these SNAP 4000 units, guess what happens? Yup, you got it, things puke and drives disappear or are not seen at all (weird crap happens). Now there were some key words there above. "With standard cables". The -003 and -004 for the most part use the same controllers and quite often even the same revision main boards, so how do they get away with doing C/S then when the -001 and -002 won't (with modern drives)? Damn good question, and I got your answer. They don't use a standard cable. If you look close at the cable on a -003 or -004 unit, you will notice that there is a cable lead cut going to the second drive on the chain. This little removal of that one elad makes all the difference in the world to the controller and hard disk(s), again, for reasons I won't go into here. So, the FIX for a -001 or -002 revision SNAP 4000 so it won't have this little bug/problem is to use C/S and to remove that one little lead on the ATA cables. That simple. And trust me, it is not hard at all. So bottom line is that you can use 4 x 250 GB drives, and do it with any revision SNAP 4000, so long as you do the proper things needed. Like update the OS if needed and change the drive configuration if it is a revision -001 or -002. Okay, David also mentioned the Power Supply. While I have been chastized for saying this before, it is still true and I will repeat it here. The SNAP 4000 power supply is just not designed for 4 modern larger drives. It is not so much the issue of running the drives, it is the issue of spinning them up on initial power up (the first couple seconds). According to spec, a Seagate 250 GB drive needs 12V @ 2.8A during spin up. Multiply that times 4 and you have 11.2A. Last time I checked, the SNAP 4000 power supply is rated at 6A on the 12V rail. You do the math. Now, in fact, the SNAP 4000 does not have a weak power supply, it has a damn good power supply. Why? Becasue even under those peak loads, a SNAP 4000 power supply will usually handle the momentary load and do the job. BUT, and you knew there had to be a but, these SNAP 4000 units are getting older by the day, so it is not uncommon to have a power supply that just can't handle that anymore. It happens with age on power supplies. So be prepared for it. By the way, replacement power supplies for the SNAP 4000 are hard to come by and are expensive when you do find them. I have come up with a MOD where you can use an inexpensive AT or ATX power supply as a replacement, but you need to know how to solder, understand wiring diagrams etc, and have a cutting tool like a dremel. It is not for the just anyone to do. Northwest Technical does have a replacement power supply they sell for $150 at http://www.northwesttechnical.com/ If you use larger drives, you will most likely need to upgrade the memory to 128 MB if it's not already, especially if you load JVM. If you don't, when you try to tansfer large amounts of continuous flow data (large files), you will get errors and the transfers can/will fail. There, now you should be armed with the knowledge you need to make a 1000 GB (1 TB) SNAP 4000. If you have questions or need help, just ask... I have 6 SNAP 4000 here and have done a ton of testing on them. Chances are, if it's hardare on a 4000 (and even on the 4100), I have seen it already. And if I don't know, I am sure David, Sam, Jontz, Hallis, or one of the other many good people here will. Note: I am a Hardware guy, not an IT, networking, or security guy. Hey David, can you tell I took my happy pills today? Last edited by Phoenix32; 03-02-2007 at 01:42 PM.