Snap 705N/4100 lives and thrives!

[soundidea]

The tip David sent re: enabling the non oem os via debug worked and the server accepted the Quantum Snap OS. The new (used) 200GB drives were recognized and formated to 130 GB each. The RAID 5 array was created properly and passed the diagnostics. Explorer shows 404 GB for the array.

I still had a problem in that the 4th drive then failed upon re-start. It would clean up and reformat eventually once you deleted the array, and reformatted the drives, but issue an error again when you re-started the server and only show 3 drives. I updated the os to
the latest patch from Adaptec, 4.0.860, and then the 4th drive registered properly again. It is all working fine now and I have 400 GB of storage for backup and archiving for my network running PCs and Macs.

I installed a Kingston 256MB PC133 dimm and it worked fine. I didn't run any comparative throughput tests before and after, but it does seem snappier in browsing and use. It is faster than the 200GB Snap1000 I also have.

I would like to quiet all the fans, but the P/S form factor is the problem. I am temped to repackage the top metal lid and use a standard, quiet PC P/S. But, a visual and mechanical cludge. For now, I just keep it off until needed.

I think the RAID 5 is a great idea. One drive can fail and be replaced and the Snap will not loose any data! I don't need to manage multiple backups anymore!

Question for the group..... Until some clever person is able to extend the drive limit of 137GB on the 4100/705N, what is the best Snap choice to hold 4 BIG drives?

Thanks again to David and the group.

Nelson
Soundidea

Snap 1000, 200GB
Snap/Dell 705N 400 GB in a RAID 5 4-disk array

[jontz]

For the money, probably a Snap 4000. As long as you are running .805 or above it will handle large drives, even the new 750GB drives.

[rpmurray]

Originally Posted by jontz

For the money, probably a Snap 4000. As long as you are running .805 or above it will handle large drives, even the new 750GB drives.

I wouldn't be too sure about that. The Seagate 500GB drives draw a bit too much power on startup for the Snap 4000 power supply. Three will work, with four it comes on for a couple of seconds and then shuts itself down.

It might work with the Hitachis since they seem to have lower startup power requirements, but they use a bit more when running than the Seagates. I don't have any Hitachis to test this with.

Right now I'm trying to replace the power supply with a 250W unit, but it's going to require some rewiring to hook it to the motherboard, and I've been tied up at work for the last few weeks, so haven't gotten very far.

[jontz]

That's intertesting, because Seagate lists every drive in their .10 series, from 200 to 750 GB, as drawing the same amount of power (13.0W operation, 12.6W seek, 9.3W idle). I suspect that this is the case with most manufacturers. There isn't any reason why there should be a wild power swing (more than a watt or so) between different capacity drives.

[blue68f100]

The only difference should be on startup, due to more mass (platters). Heavier platters require more power to get them spinning.

[Phoenix32]

David made a good point there. The mass of the platters will make a difference in start up power needed from model to model. There will also be a small difference in normal operation (for the same reason), but it will be by far much less significant than is the start up power needed. Most manufacturers post spec requirements for a particular series of drives, using the highest requirements for a given model within that series. I would bet that even though the drives of a series may all show as the same spec requirements, that in fact, the power drain for each drive in that series goes up as the number of platters increases.

This does bring up a good question, one which I had intended to ask. Just how much can a 4000 (or 4100) handle? I have seen mixed information here on the forum as to what they can handle. I have seen some say (4) 250 Gig drives is no problem for a 4000 (and see them for sale that way on Ebay also), but I have also seen others say they had problems with 4 drives (unknown to me the size at this moment). I know one company in Oregon who sells 4000's with 4 Drives, and I believe they are Western Digital 250 and 300/320 (not sure), but not known which model.

I do not recall seeing that type of info in the Wiki when it was up, so maybe this would be a good time to ask people what they have had success and failures with in the 4000 and 4100 Snaps Servers. How many drives, what size, what brands, what models, etc... Maybe a good baseline can come of it and be added to the Wiki later on?

I myself have a 4000 and have yet to have more than 2 drives (250 Gig WD) in mine. I do plan on putting 4 drives in as soon as I can determine the information asked above. I would bet others have had this question as well.


My apologies to the Forum Admin if I have hijacked this thread and it needs to be made into a new topic.

[jontz]

Good thoughts about the start up power surge. The 500GB Seagates do have four platters to get moving, the Hitachi's have five. That is a fair amount of mass to get moving. I wonder if adding some relatively large value capacitors to the output of the power supply could help? I don't have a 4000, so I take it that it attempts to start up all four drives at once? My 4100 starts up drive one, then drive two, etc, to avoid the power supply shutdown problem that you are having with your 4000. The irony here is that the 4100 can't handle large drives because of the LBA48 bit issue!

I have had my 4100 up and going with a single WD 1600JB 160 GB drive, recognized as 137GB, formatted to 128GB. I was planning on getting three more of these drives and just letting the rest of the space be wasted, because I would still be getting 17GB more per drive vs using 120's. Simple economics made my decision though when I get a screaming deal on four brand new Seagate 120GBs. I am running a 4 disk RAID 5 array with no problems.

[rpmurray]

All I can tell you is what my 4000 has been able to do so far. With 3 Seagate Barracudas (ST3500641A) I can get it to boot and stay up. With 4 it will shut itself off after a few seconds.

From the Storage Review website I got the following numbers.

Hitachi Deskstar (500 GB), Idle 8.5 watts, Active 14.6 watts, Peak 17.9 watts (12V) 4.4 watts (5V).

Seagate Barracuda (500 GB), Idle 8.6 watts, Active 12.6 watts, Peak 29.4 watts (12V) 3.5 watts (5V).

Seagate Barracuda (250 GB), Idle 6.7 watts, Active 10.6 watts, Peak 26.4 watts (12V) 3.5 watts (5V).

I believe some folks have been able to get four Seagate 250s to run in the 4000, which leads be to believe that the 12W difference in Peak on startup is enough of a hump to keep it from staying on. It seems to startup all four drives at once.

[blue68f100]

I will need to check, Hitachi or Seagate on their advanced utility Disk (downloaded from their site) had an option for a delay start. I ran across this 6 mo ago when I had a drive acting up. Just don't remember who. If I have time I will dig them out and check. I think it's Hitachi.

I dropped Seagate and Hitachi an email to see if such a utility exist. And/or is it part of the SMART technology.

[blue68f100]

Ok,

Received resonse from Seagate and Hitachi.

Seagate said NO on PATA drives there is something for SATA.

Hitachi Said YES ......BUT .... It must receive a signal from the controller (BIOS or Software) to start up. Here is the link for jumper setup if someone has Hitachi drives installed to test. http://www.hgst.com/hdd/support/7k400/7k400jum.htm

[Phoenix32]

I did a some searching on the Seagate and Western Digital factory websites. As I suggested before, they are all using a pat standard for each drive series on their power requirements. In fact, Western Digital does not even list spin up power on some of the drives here and there. One set showed the same power requirements for an 80 Gig drive as it's 500 Gig drive in the same series. We know that can't be so. I would tend to agree with what rpmurray said about that extra 12 watts being the division line. The problem is getting usable real data on the spin up requirements for given hard drives. Storage review seems to be a good source to look, but I cannot get on their forum. Seems my e-mail address is already in use, but I have not been there in a long time and cannot remember my password and my username does not seem to be listed (beats me).

With no ability to start the drives one at a time or delay them it seems, anyone else got experience with what worked and didn't work in the 4000 for Hard Drives?

[blue68f100]

rpmurray show the Hitachi being 10-12 watts lower than the seagates. At just under 18 watts x 4 = 72w + MB for a 500gig drive.

The 4000 has what 120w ps?

Where 4 seagates are pushing 120w.

[Phoenix32]

Let's not forget the 5volt rail in that wattage equation.

Also, the power supply is a 120 watt power supply, but what is it rated at on just the 12 volt rail (where all the spin up power comes from)?

Then... While the power supply might be rated at 120 Watts (power supply ratings are always a bit speculative), but that is supposed to be continuous, not peak. What is the power supply rated at for peak load, on th 12 volt rail?

I swear, I am not trying to be a smart ares here, promose. My point is that determining the power we can use is not as simple as we have put it here. To do it that way, we would need to know the actual peak load rating/capability of the power supply on the 12 volt rail. Our best bet is to go by examples of what has worked and what has not worked for others, -OR- find out these numbers (actual peak rating, and actual spin up loads).

[K3nt]

You might also want to take into consideration the Amps these power supllies put out. Sometimes the watts are not that interesting and the combined amp requirement. If the drives require 1.5amps each to power up and the power supply can only manage 5 amps on the 5v and 12v lines then 3 drives would be ok, but not 4.

Just a thought.

[rpmurray]

The only information I've been able to find out about the power supply in the 4000 is that it's a 3P Pacific Power Products Model No PRO 120-3. It produces +5V 12.0 A; +12V 6.0 A; -12V 0.5 A.

The power supply in the 4100 is an EMACS MPW-6150F. It produces +5V 10.0 A; +12V 6.0 A; -12V 1.0 A.

Startup current for the Seagates 500GB is 2.6 amps, 11 secs from power on to ready. Looks like the SATAs draw an extra 0.2 amps on startup.

[jontz]

So really the only foolproof way to get this to work is to beef up the power supply, either by putting in a new one or by possibly adding some capacitors to aid in supplying startup current. It would be nice to be able to do a delay start on two of the drives, but it doesn't sound like anyone has had any luck finding out if Seagate or Hitachi support this.

This brings out another intertesting point. The 4100 starts up each drive one at a time, but according to rpmurray the 4100 power supply supplys 2 less amps on the 12V rail than the 4000. Even if someone cracks the LBA48 bit issue, I wonder if the power supply can handle large drives, even after spinup.....

[Phoenix32]

Originally Posted by rpmurray

The only information I've been able to find out about the power supply in the 4000 is that it's a 3P Pacific Power Products Model No PRO 120-3. It produces +5V 12.0 A; +12V 6.0 A; -12V 0.5 A.

The power supply in the 4100 is an EMACS MPW-6150F. It produces +5V 10.0 A; +12V 6.0 A; -12V 1.0 A.

Startup current for the Seagates 500GB is 2.6 amps, 11 secs from power on to ready. Looks like the SATAs draw an extra 0.2 amps on startup.

This would seem to indicate that with the Seagate 500's we would need 10.4 A; for 11 seconds; PEAK power, from the 4000 Power Supply. We know (I think) that this doesn't work, so we know the 4000 PS wil not deliver that much for that long.

It would also indicate that the 4000 PS can deleiver 8.8 A; for 11 seconds; PEAK power. This is derived from the 26.4 watts (2.2 A) each from the Seagate 250 drives that seem to work. 2.2 A x 4 = 8.8 A

We are bracketing it here. At least 8.8A but not 10.4 A. This is why I was asking if anyone had more data on what worked and didn't (we might be able to bracket it further).

[Phoenix32]

Originally Posted by jontz

So really the only foolproof way to get this to work is to beef up the power supply, either by putting in a new one or by possibly adding some capacitors to aid in supplying startup current. It would be nice to be able to do a delay start on two of the drives, but it doesn't sound like anyone has had any luck finding out if Seagate or Hitachi support this.

This brings out another intertesting point. The 4100 starts up each drive one at a time, but according to rpmurray the 4100 power supply supplys 2 less amps on the 12V rail than the 4000. Even if someone cracks the LBA48 bit issue, I wonder if the power supply can handle large drives, even after spinup.....

I don't think adding Caps to the power supply will help. 11 Seconds is a long time for a Cap to assist. -AND- Those caps still need to be charged. Which means they would draw more current to charge at the crucial time we need all the current to spin up our drives, thus having the reverse effect. Maybe not, but this is what I suspect.

As for the 4100, 1.5 A; each for the drives in a 4100 for normal operation is plenty I would think. No drives I know of draw that much current while idle or even accessing; it is the spin up that is the killer. Sure, the 4th drive that is powering up might draw 2.6 A; or say a larger drive drawing 2.8 A; during spin up, leaving only 6 A - 2.8 A = 3.2 A for the other 3 spun up drives; 3.2 A / 3 Drives = 1.07 A for each drive. I think this is enough for most any ATA drive I am aware of for normal idle or accessing. If not, remember, even that power supply has a little extra in reserve for PEAK load. In other words, keep cracking away on that LBA48 solution.

[blue68f100]

If you put a line switch on 1-2 of the drives. You could manually stage them. Just make sure you have a good UPS unit. On startup the OS will proably report a bad drives (2) then you have problems with raid 5. It will proably start up with 3, then have the spare drive switched. The when a problem accours turn it on. Not running all the time in stand by you will have a good drive when that accours. We all know the Snap can rebuild an array with 3 drives, so if one starts up late, it will work. But will have the long resync or file check.

I think the in rush only last till the drives start spinning from a dead stop. Then drops off pretty quick.

If you have a patch or extender cable. Setup a fluke to measure the inrush, then you will have accurate data. Another note. My 2000 (fried) had a problem with loose power connectors early on. It tripped the fuse. After cleaning the connectors and tightening the sockets everything worked till the fan failed.

Just my 2 cents.

Not having one to play with, your on your own.

[jontz]

Originally Posted by Phoenix32

I don't think adding Caps to the power supply will help. 11 Seconds is a long time for a Cap to assist. -AND- Those caps still need to be charged. Which means they would draw more current to charge at the crucial time we need all the current to spin up our drives, thus having the reverse effect. Maybe not, but this is what I suspect.

As for the 4100, 1.5 A; each for the drives in a 4100 for normal operation is plenty I would think. No drives I know of draw that much current while idle or even accessing; it is the spin up that is the killer. Sure, the 4th drive that is powering up might draw 2.6 A; or say a larger drive drawing 2.8 A; during spin up, leaving only 6 A - 2.8 A = 3.2 A for the other 3 spun up drives; 3.2 A / 3 Drives = 1.07 A for each drive. I think this is enough for most any ATA drive I am aware of for normal idle or accessing. If not, remember, even that power supply has a little extra in reserve for PEAK load. In other words, keep cracking away on that LBA48 solution.

Ideally you would charge the caps after install and before you power up the unit by detaching the drives and using a resister inline so that the caps don't suck all the power at once and fry the power supply. After the caps are charged, plug the drives in and start up as normal. The caps supply the startup current and then recharge when the drives stop drawing as much power, just like a car audio cap on a subwoofer amp. I can't imagine that these drives draw peak current for 11 seconds. It doesn't take a drive that long to spin up. I am thinking 2-3 seconds, tops.

True about the 4100 power supply. I just hate "barely adequate" power solutions. Seems like it would work, and I was in no way telling people not to fix the LBA 48 bit issue, trust me. I would love to have 4 750 GB drives in my unit

[Phoenix32]

Originally Posted by jontz

Ideally you would charge the caps after install and before you power up the unit by detaching the drives and using a resister inline so that the caps don't suck all the power at once and fry the power supply. After the caps are charged, plug the drives in and start up as normal. The caps supply the startup current and then recharge when the drives stop drawing as much power, just like a car audio cap on a subwoofer amp.

Me thinks it would just be easier to change out the power supply.

Originally Posted by jontz

I can't imagine that these drives draw peak current for 11 seconds. It doesn't take a drive that long to spin up. I am thinking 2-3 seconds, tops.

I don't think they draw that much power for 11 seconds either, but I was going on what rpmurray said, which I think is what the spec says.

Originally Posted by jontz

True about the 4100 power supply. I just hate "barely adequate" power solutions.

Agreed, in fact if you expect to get much life from a power supply, you should run it at about 75% of rated spec or less. Just like any other electronic component(s), running at maximum spec shortens life a lot.

Originally Posted by jontz

Seems like it would work, and I was in no way telling people not to fix the LBA 48 bit issue, trust me. I would love to have 4 750 GB drives in my unit

LOL, yeah, me too! But who can afford it...

[Phoenix32]

Originally Posted by blue68f100

If you put a line switch on 1-2 of the drives. You could manually stage them. Just make sure you have a good UPS unit. On startup the OS will proably report a bad drives (2) then you have problems with raid 5. It will proably start up with 3, then have the spare drive switched. The when a problem accours turn it on. Not running all the time in stand by you will have a good drive when that accours. We all know the Snap can rebuild an array with 3 drives, so if one starts up late, it will work. But will have the long resync or file check.

I think we are now back to, "it would be easier to just change the power supply".

Originally Posted by blue68f100

If you have a patch or extender cable. Setup a fluke to measure the inrush, then you will have accurate data. Another note. My 2000 (fried) had a problem with loose power connectors early on. It tripped the fuse. After cleaning the connectors and tightening the sockets everything worked till the fan failed.

Both good ideas.

Originally Posted by blue68f100

Just my 2 cents.

Not having one to play with, your on your own.

May be, but you are still a LOT of help...

[blue68f100]

OK, Here is the rest of the nickle.

Power up the drives 2 at a time, then the controller board.

[rpmurray]

I'd love to see someone try the Hitachi drives as a solution, since they seem to draw less power on startup. At the moment I can't afford to get four of them to try it out.

[jontz]

Originally Posted by Phoenix32

Me thinks it would just be easier to change out the power supply.

Probably, but what fun is that? And since when are we interested in doing things the EASY way around here?

Originally Posted by Phoenix32

Agreed, in fact if you expect to get much life from a power supply, you should run it at about 75% of rated spec or less. Just like any other electronic component(s), running at maximum spec shortens life a lot.

Yep. Probably part of the master scheme though. I'm sure adaptec doesn't want these things up and going for too long before you have to buy a new one.

Originally Posted by Phoenix32

LOL, yeah, me too! But who can afford it...

If everyone in this forum wants to send me $10 or so, I'll let you know if it works