[Phoenix32]

Originally Posted by MarlboroMan

I have 2 of the Snap Server 4000 series units. I have examined this thread and the messages in it at great length and I am a bit confused on a few things that have been said here, I hope that Phoenix32 can clear up some of the confusion for me.

Well then, let's see if we can sort this out step by step...

Originally Posted by MarlboroMan

I upgraded my first 4000 from 4x30GB to 4X250GB drives. The problem that I continue to have is that once the new drives are installed, the unit goes into disk check mode and when it gets done with that, it just works fine till I copy a few files to it and then it just reboots. I have the latest version of the SNAP OS version 4.0.860 if that matters.

This could be a number of things. Bad power supply, bad board, bad memory, bad OS load, etc... In other words, could be tough to sort out without more info, but let's continue since I think you cover more of it in your message.

Originally Posted by MarlboroMan

My best guess based on the info here is now that I am going to be in need of both the power supply upgrade and the RAM upgrade to get the unit running.

Either or both of those very likely candidates. If the power supply cannot handle the load of the drives during spin up, the SNAP 4000 can do all kinds of mysterious things that are difficult to track down. Usually, one or more drives will drop in and out as orphans to a RAID or say they are not present as a single share or whatever during boot up, but the symptoms can vary and be quite annoying.

I have found through trial and error that with the larger drives, especially if you are running JVM which will make matters even worse, you need 128 MB of memory for it to function properly. The normal problem will be that it will just cancel/abort large file transfers. I have not seen it cause a reboot, but this does not mean it couldn't.

Originally Posted by MarlboroMan

What is your reccomendation as far as the RAM upgrade? What is the max RAM that you can put in it? What type of RAM specifically does it take or what works best?

I recommend 128 MB of memory for a SNAP 4000 (and 4100) if you are going to use larger hard disks, JVM, or both. For simple home use, I have not found any advantage of going to 256 MB of memory during any of my testing. However, if the SNAP server will be under heavy loads (many users), using lots of JVM stuff, or things like that, then 256 MB of memory will be needed, or at least be of some help.

The maximum memory limit for the SNAP 4000 (reguardless of what one bozo keeps advertising on his auctions on eBay) is 256 MB. The chipset will support more, but the configuration of the board only allows 256 MB.

The SNAP 4000 uses Low Density PC 100 SDRAM. It can use High Density memory, but you will only get half of the memory on the DIMM. It can also use PC 133 memory, but most PC 133 memory is High Density. This does not mean PC 100 memory is Low Density either. There is a lot of High Density PC 100 out there also. BE SURE what you are getting before you buy it.

As for what works best, I have used various memory modules in my SNAP Servers and have no real preference. However, if you do a search here on the forum, some peple have posted memory brand and models numbers that have worked for them. The 4100 and 4000 use the same memory, so it is interchangable. The difference is the height. 4100's need lower profile memory to clear the casing.

Originally Posted by MarlboroMan

The PSU question seems to be much more of a major issue than I first thought that it might be. It looks like the Northwest Technical solution for the PSU replacement looks good, but the $150 per PSU price tag is kind of steep with all things considered. I am not sure that I want to put $300 in these two Snap servers.

Okay, this part of the message is going to get longer, but because I have posted this information on several occasions, I will try to keep it short and direct so as not to lengthen this message too much.

First, do not read this as a negative towards the NW Tech power supply. It is a good solution for what it was inteneded for. Yes, it is steep for price. But, it was intended for corp IT usage where time and easy installation were paramount and money not the issue. Their power supply is designed so that pretty much anyone can install it and not take a lot of time. The down sides are that it does require one small modification on the switch, it does not provide any more power than the original stock OEM power supply did, and it is expensive. Please take note that I said IT DOES NOT PROVIDE ANY MORE POWER THAN AN ORIGINAL STOCK OEM POWER SUPPLY. If you are a corporate user with a down SNAP 4000 or a home user who is desperate, this power supply is the answer and damn good to have available. If you are a home user on a small budget, then it may not be for you. Only you can answer that one.

Here is the short version of the issue. Larger hard disk take more power to spin them up initially (the first few seconds of each power up). This can be to the tune of 2.8A @ 12V per drive for a Seagate 250 GB drive as an example. So with 4 drives, during that initial spin up, it can take upwards of 11.2A @12V for that first few seconds. After initial spin up, it drops to around 3A to 4.5A @12V for those same 4 drives (depending on the drives being used). The SNAP 4000 power supply is only rated at 6A @ 12V. You can see the problem here I hope.

With that said, the SNAP 4000 OEM power supply was a decent power supply and usually can handle those momentary surges during spin up. Good on Quantum for that one (Notice I gave credit to Quantum, not Adaptec). But there is a down side here too. The SNAP 4000 units have been around a while and are getting older every day. Power supplies weaken with age and use. Some of the SNAP 4000 power supplies out there are getting weaker now and starting to fail, or will fail when put under the heavier load of larger drives. It just is what it is.

If you have this problem, the solution(s) are this. Get a new power supply, if you can find one (and you will pay dearly for it), get another used one and hope it is better/stronger, buy the one from NW Tech (keeping in mind it is not rated for any more than the stock OEM power supply was), or doing a power supply MOD yourself. I designed a power supply MOD a while back and it is documented elsewhere on this forum, but it is not for just anyone to do. You need to know how to solder, can read wiring diagrams, and have the tools for doing it. It is not really hard for a person who knows these things (pretty easy in fact), it is just time consuming, but for someone who doesn't know these things, it is just not a very good idea. Your only other option would be to get someone to do one for you. I have considered it, but right now my time is tied up pretty good, so it would take a while.

I know these may not be the answers you wanted to hear, but they are the answers. If you find other solutions, please share them, but with as many people as who have looked, I doubt it, but who knows.

Originally Posted by MarlboroMan

As for the hard drives, I have 4 Western Digital 250GB JB series drives to go into one unit and 4 160GB Seagate series 9 drives to go into the other unit. Do I have to have the replacement PSU to handle these drives?

I hope I answered this above. The answer is, it depends. Depends on how good your power supplies are (how strong or weak), which I cannot tell you here sitting on a forum. Each one is a little different.

Originally Posted by MarlboroMan

I am also confused about the entire cable/master/slave/Cable select issues. Which mode should I be using?? Can you just replace the cable?? Why could you not just lock the hard drives down to a specific ATA mode to make it more compatible with the SNAP server? Western Digital does offer such a utility for use with their drives which allows you to lock the drive to specific ATA mode? Which mode does the 4000 series use??? I need some additional help understanding this issue on the impact of this on the performance of the 4000.

For -001 and -002 revision models, they use Master/Slave. For -003 and -004 revisions, they use C/S.

You should use C/S if at all possible, or you may run into the bug described earlier if/when you have a failed drive, and lose your RAID 5 array, and thus all your data.

To use the C/S, you need to use a cable specific for this purpose, which came with the -003 and -004 models but not -001 and -002 models. The cabes from a -001 or -002 model can be modified easily. It just requires removing/cutting one lead.

Yes, you could lock the drive to a specific ATA mode, and this is always a good idea. WD does have a utility for this, but most other drives, like Seagte do not. I think Hitachi does also, not sure, but none of the others do that I am aware of.

When I write this information, I have to provide it for the largest group of people. -AND- I have not tested it with drives with the ATA set and cannot guaruntee the results. If you have large hard disks, with lots of your precious data on them, do you really want to risk it in hopes that it will be fine just because you locked the ATA mode, rather than make a simple MOD to your cables (if needed)? See my point? Put another way, this was a known problem to Quantum (and now Adaptec) and they resolved it with a cable modification. If their engineers felt this was the way to go, it was good enough for me.

Plus, I could never PROVE FOR FACT it was the ATA mode causing the problem for sure in C/S without the modified cable(s). I spoke with David at great length about this, but neither of us could prove it, but we both felt it was the problem. We are pretty sure, but can't prove it as hard cold fact. I can say for fact though, that without the modified cable, none of the drives I tested (several brands and models) would work in C/S mode other than the old Quantum 30 GB drives.

I did consider setting the ATA mode with a utility, but knowing this would not be a fix for most people, and considering how long it would take to run the tests, I didn't do it. Consider I would have to do it a number of times to be sure. Which would include, setting the ATA mode, formatting the drives, building the array, then failing a drive (and you have to try it in different drive positions to prove it), then see what happens, and let it rebuild, -if- it will rebuild. Say do this at least three times in each drive position to verify results because it doesn't always happen. Now keep in mind how long it takes to build or rebuild those arrays with large hard disks. You do the math. It would be a lot of time and work to prove out a method that MIGHT work for a minority of people and there is an easier solution available that engineers thought was best. Would you have done it, or wrote it up any differently? I am not trying to chastise you here, I am just trying tto help you understand why it is written as it is written. But note that I did give tthe information so that someone like yourself could try it if they wanted to.

Make more sense now?

As for what the issue is and the impact of what is has on the 4000, I thought I wrote all this up before several times, but I will give a short answer here again.

In the Master/Slave configuration, if/when you have a hard disk fail, if you power the unit without replacing the bad drive, or the drive fails in a manner that will keep the IDE controller from seeing the drivel, there is a very strong possibility that the SNAP 4000 will fail/orphan the other drive on that IDE controller (the paired drive). If this happens, then the RAID 5 has lost 2 drives now. If you lose 2 drives in a RAID 5, the array is broken and your data is gone (without some specialty recovery services). I would say that is a pretty big impact.

Originally Posted by MarlboroMan

In the end, I would like to just use the two SNAP sservers as RAID 5 backup and file storage boxes. Traffic on them should not be that heavy.

To that end, I hope I have answered your questions here and pointed you in the right direction(s). If not, keep asking and I/we will try to answer as best we can. Just remember, we are not paid tech support from Adaptec.

Originally Posted by MarlboroMan

Would like to hear all suggestions on the best way to tackle this project and get them stable like my SNAP server 1100 is with the 320GB Western Digital hard drive in it.

I think I gave some suggestions here and there throughout this message thread and this mesage. But, here are a few more.

- For your power supply question, the only thing I can suggest at the moment is to remove/disconnect (from the IDE and power), one drive, then two, etc... See if the unit starts working properly. If it does, that is a good indication the power supply is too weak to hanle all the drives.

- For the memory, maybe you have a bad stick of memory. Replace it with a different stick and see what happens. Preferably with a 128 MB DIMM.

- For the OS, go back to a KNOWN GOOD configuration, even if this means with one drive. And be 100% sure it is a good configuration (test things). Then do another OS update (reflash the OS). DO NOT do an OS flash if the system is not operating properly. You run the risk of corrupting the OS, which may be the problem here now.


I hope this helps. It took me long enough to type up.

David (or others), do you agree or disagree with what I have said here? Do you have anything to add? The more the better...