Powerloss, 4400 crashed, RAID 5 d:\ missing lights green 1

[awaywifye]

Tried to make it as compact as possible.
About 6 months ago one of our sites abruptly lost power. The 4400 crashed, and they lost the container (RAID 5 - 3x18GB HDD). After about a week we had the sight completely up and running (our servers are imaged - only this site hadn't imaged their server for about 3 years forcing us to rebuild the server and network from scratch - 1 stand alone server with x computers).

Just today hey took another shot (destroyed the AC at the site) and the server went down again. After some discussion with the guys on site, the D:\ (RAID 5) is gone. All green lights, recognizes that there is a container there and displays it's size, but we cant see or touch it.

Want to try and spare these guys another rebuild (they have recent images and tape backups, since it's the d:\ they won't have to rebuild their network). Any suggestions?

 

[lastdonuk]

Hello awaywifye,
I'm sorry I don't have any suggestions for you, but one of our 4600s crashed on reboot caused by a power surge this morning and I'm just wondering if you had had any suggestions given to you since you posted this? Apologies for not helping with your problem, but it seems the issues are similar - we get blue screened on startup of W2K along with a heap of previous messages along the lines of "Missing Container" in RAID-5, etc...

Thanks anyway, sorry if I've wasted yr time.

 

[awaywifye]

How is your server set up?
All of our servers are RAID-0 (2 drives) for the C: and RAID-5 (3 drives) for the D:

The lab wasn't receiving an error about a missing container, in fact upon startup the container and size are recognized. When they went into the Array manager, drive 2 of 3 was bad. We tried reseating the HDD (in the blind hope that it would rebuild itself), we tried booting up with drive 2 out but were still unable to see to D: once windows loaded. We even tried using Plextor's driveimage and seeing if it would recognize a d: for imaging.
So far the lab has been on the phone with Dell on and off for about a week. They originally repaced drive #2. This did not solve the problem, so Dell sent out another HDD and told the lab to replace drive #3 as well.

We did run into a similar problem to yours once (the missing container). For us, all drives were good, but the container became corrupted following a power surge. We had to completely rebuild the container which killed EVERYTHING on the D:. Since the lab didn't image their server for almost 2 years (no one knew the admin account or password from 2 years ago), we had to completely rebuild their network from scratch.

I don't know if this is any help or not, but go into the array manager (Should be ctrl+M or ctrl+A) upon reboot. What does it say there? Drive bad, container bad?

 

[lastdonuk]

Thanks for the prompt reply awaywifye. We have a raid 5 (6 drives.) The array manager states that the container is in 'critical status' Also shows that number 5 is a missing member all other discs can be seen. There is no fault light showing on this drive number which is strange. Also states when booting up that 'Container has missing members and are degraded'. It then proceeds to the windows 2000 server screen and then displays blue screen with stop error 0x0000007b Inaccesable_boot_device. I have tried starting in safe mode and last good known configuration but to no joy. Any further help would be appreciated??

 

[awaywifye]

Ok, sounds like yor #5 took a dive on you. One last thing to try would be to reseat the hard drive, if you're lucky, the hard drive will be rebuilt and you can continue on your way. If this doesn't work, and you have a dell contract, you'll need to call them and get another hard drive sent out.

Good news:
Container sounds like it's salvageable = no data lost
Bad news:
#5 is dead = need new hdd = time lost = possible money lost.

 

[lastdonuk]

thanks for your help and advice, awaywifye - and sorry for the delay in the thanks.

the machine is unsaveable, but somehow we managed to get all the data off it by using the W2K repair facility on the boot disk.

anyway, thanks again - we all learnt something and we've all survived the experience.

cheers!

 

[technome]

My question is why are you using raid 0.. this something you do not do on a production server, under any circumstances, for the OS or data . The only place Raid 0 is warranted, is for a partition holding the pagefile and temp files, imaged or not. If any of you servers are using "dynamic" disks, they should not be.

Something is seriously wrong if you lose arrays like this. I have had power failures, pulled plug, brown out, power button shutdowns etc >300 times, and have never lost an array as a result in 15 years.
I divide my arrays into two partition, one for the OS and one for data; I have had to rebuilds the OS partition a few times but the data partitions are not affected by the OS partition.

 

[NickFerrar]

I think (hope) he probably meant RAID-1 for the OS.
I'd agree that RAID systems should handle a failed disk without issues, that is what they're designed for after all. We've lost plenty of drives in arrays and never lost any data yet.

 

[awaywifye]

NickFerrar, yes sorry I meant RAID-1. Never even noticed that.

Lastdonuk glad everything worked out in the end. Sorry I couldn't of been more helpful.

technome, thank you for your response, however you jumping in with no advice for the issue doesn't help out any. Something is not seriously wrong. It's called working for the military.
The power surge that took out one of our servers also blew out an entire AC unit and power units in several other pieces of gear. Our servers run on generators, they're exposed to the elements, they've travelled more miles then you ever will. So I'm happy that in your controlled environment you've never had any problems, but we have.

 

[technome]

Sorry I was bit rough there, there are so many "techs" doing raid 0 whith massive losses, I see it in posts all the time.
The military.. I can see the problem. I worked a couple of times at a nearby base, and I do not know how the techs keep things together; in both situations the techs had so little to work with.