SNAP4500 Not Booting - Begging for some help!

[SilentRaven]

Okay here is the situation. We have a SNAP4500 with 1Tb storage (4x250Gb IDE) that suffered an abrupt power outage the other day. When I rebooted the system it hangs at "Decompressing Linux.. OK, Booting the Kernel" - then nothing.

I'm not very familiar with these devices and have attempted some basic T/S.

I've reseated all the drives - no change

I've removed all the drives - which stops the system from booting (No OS Found)

I've tried swapping drives (ie moving drive3 to drive0 and removing the others) - no change

I've tried accessing the user web interface (unable to ping the IP address)

Tried resetting the defaults - no change.

As it stands, I'm lost as how to proceed. We have a duplicate device in Knoxville, TN - but that's an 8hr drive - but it sounds like the only way I can rebuild the GSU. If so, can someone please explain the exact steps that would be involved in making that work. Would it be as simple as swapping out the bad GSU drive with a good one and letting it rebuild across the RAID?

Again any help is appreciated!

Thanks,
Ben

[blue68f100]

You have to be very careful swaping drives around that make up a Raid array. This can and will break the arrays and data may be lost. Changing the order will also loose all of your data. No UPS to protect servers is not good practice. If not already numbered, number your drives.

But your 4500 indicates that it did not make it that far. I would connect a null modem cable to the com port and watch the boot process in detail from a terminal. Not sure of the settings for terminal, thinking around 119.2k N81. You are not booting far enough to trigger disaster recovery. If you were you could reload the OS.

Are the fans and blowers running and at full speed on power-up?
If left on does the watchdog kicks in and attempt to reboot to another drive? Normally this timer is set for 5min.

If you want to get a working drive. To do it right you will force your working unit to do a raid rebuild. This works best if AV, SnapShots, BU, and any other process that keeps up with drive info MUST be turned OFF.
Power down the unit.
Remove one drive ( from slots 2-4)
Install new HD in removed drive
Boot server, in about 5 min you will see drive activity on the new HD (max 8-10min)
Once completed power down and remove the freshly loaded HD
Install original HD and reboot. If the 4500 does not resync, you will need to force it. ( or reboot and hot swap the Orig HD, and resync will happen automaticly.

Now install the freshly loaded HD into Slot 1 of your failed 4500 and boot the server. If all is well it will boot and start a resync. But at this point I'm not sure I would trust the array, since it was unable to find a good drive to boot off of. I would retrive the data needed and then wipe the units and start over from scratch.

[SilentRaven]

Thanks for the quick reply. But I'm thinking at this point we to just decom this device and move the data over to our SAN. Sounds like these things are far more work then they are worth

[blue68f100]

Normally these units are rock solid and have very few issues. When I got mine I did extensive testing to see how they handled power loss. What I found was I could not get it to fail doing writes and read by pulling the power in the middle of the task. But you need to be running snapshots for the unit to self repair.

Now it's possiable that the surge/outage may have done some hardware damage. I would have your other office send you a working drive and see if it will boot. At least get you into the recovery console.

But what ever you do you need to run a UPS unit that are capable of shutting hardware down to cover power outages. No computer likes to loose its power, prefer to be shutdown properly.

[SilentRaven]

I'll talk to my supervisor about the drive, but at this point I'm pretty sure we are going to scrap the SNAP and move over to the SAN. I'm pushing to virtualize the 4 (older p3 xeon) servers anyway.

[eschw95458]

I know this might sound crazy, have you tried replacing the cmos battery??

[SilentRaven]

Not as of yet? Should we consider this?

[eschw95458]

My first question would be - Do you have the data backed up from this unit, if you do then there is no downside. If you don't it could "lose" the raid array, but I think that info gets stored on flash and not in cmos. But, yes, I have had 4500's that would not boot until I put in a new battery and cleared the cmos.

[SilentRaven]

We've migrated the data to our SAN the day after it went down.

[eschw95458]

Then there is no downside. Also if the unit won't boot, you don't need to swap drives, just start pulling drives one at a time. IE pull drive one and see if the unit will boot, If that doesn't work then leave drive one out and also pull drive 2. etc..etc. The os gets replicated across all of the disks

[SilentRaven]

I've already tried pulling the drives one by one with no luck.

[eschw95458]

I believe that particular battery is an energizer 2032, you should be able to pick it up just about anywhere for a couple of bucks, if that doesn't work you will have to see if phoenix might take a look at it for you.

[SilentRaven]

I'll give the CMOS battery a try. Never would have thought of that. I'll post my results when I get it swapped out.

[blue68f100]

Do you get any post at all, from the BIOS? can be monitored with a monitor connected. If I recall there are some settings that need to be changed if battery is replaced. Been 4 yrs since I replaced one.

[Phoenix32]

Originally Posted by eschw95458

if that doesn't work you will have to see if phoenix might take a look at it for you.

As a sound of sighs and groans creep over the entire forum

Use the evil ones name and he shall appear

-

Not knowing all the little details, only what is posted here, not being there or talking to him directly, here is my basic evaluation;

Eric, IMO the CMOS idea is reaching in this case, but -IS- possible and is worth a shot.

As David said, there may be hardware damage, but it doesn't sound like it from what I see here.

My best guess, it has a scrambled GOS. If trying to boot from each drive, one at a time, being the only drive in the unit, doesn't work, then it is scrambled across all the drives and will need a seed drive to re-install. This is not uncommon for power failures with these units.

As for the array, if I read all this correctly on what was done while troubleshooting, the array got killed by the operator and is done for. Bye bye data, sorry.

Recovery will require wiping the drives and then using a GOS seed drive to get the GOS back on the unit.

[SilentRaven]

I have to agree. My guess is that we would need a to reinstall (seed) the drives again. It's not really worth the cost to us, since we would have to pay to have a warranty reinstated on this device. Thanks for all the help guys!

[Phoenix32]

Retire it my direction, refurbishing these units is what I do.

[Max8]

Originally Posted by Phoenix32

As a sound of sighs and groans creep over the entire forum

Use the evil ones name and he shall appear

I seriously doubt that there were too many sighs and groans Phoenix!

You are a poet...You made my day with this comment!

I found it so hilarious it made me snort my coffee over my 12 year old, recently-cleaned, Heavy-ass, reclaimed from a computer fair, CompaQ keyboard...


Have a lovely Friday & an even better Weekend to all my northern hemisphere friends!

[SilentRaven]

Okay - We have migrated the services to our SAN and have retired the SnapServer. I of course claimed it and now would like to wipe it and install a different OS (I'm looking to be cheap or/preferably free). I have a copy of MS Server 2003 Standard from my school and was thinking about installing it.

I notice that when the device boots, it loads what looks to me like a SCSI adapter? Are the IDE drives RAIDed through some sort of adapter? Is there a way to remove the RAID without booting to GOS? I see says it uses SIL0680 ATA/133 Controller, but I don't see how I can interact with this chipset. Some other SIL devices offer the option of working with them via a keyboard command like CTRL+F4 and up popups a like GUI interface.

Replacing the battery did not work - but I appreciate the idea - would not have thought of that myself.

[blue68f100]

A std 4500 comes with a SCSI card for tape backup. The 4500 use a software RAID, 2 different controller chips. Most only see 1 chip. If you use a linux based OS it will see all of the hardware. There are only a few overrides that needs to be done. If I recall some one installed FreeNAS on a 4500. But if it was me I would stick with the GOS it a very proven OS.

Use the search feature and see if you can locate the threads that cover these OS Swap. But if you ever want to go back to the GOS you will need to save a working HD with the OS.

[SilentRaven]

I would love to stick with GOS, but not at the prices I've been quoted. I'd prefer to load a free application on it and keep going. Sadly the GOS is toast on this unit

Is there some way to interface with the RAID?

[blue68f100]

Here is the link to installing FreeNAS on a 4500

http://forums.procooling.com/vbb/sho...hlight=FREENAS

[Phoenix32]

When it comes to File Server/NAS, it is pretty hard to beat the GOS. The other free type NAS offerings are, well, just okay, but far from the same stability etc. That and they have not worked very well on the 4200/4500/15000 family. Sadly, for the 4200/4500/15000 series without GOS, you are better off with Win2K/2K03 storage server revisions than with the others.

[willPower]

.

[Phoenix32]

Well, now that we have a Linux guy here, maybe we can get more of these fail-to-boot units back up and running without having to reinstall the GOS!

Again, BOUT TIME! lol