Another RAID failure
Well actually it is another WD 120 SATA failure. The drives were new in october 03 and the first one started acting flakey around christmas. It took a month to get a replacement. Well here we are a few months later and the other drive is going. I went to do my weekly ghost image and half way through the C: drive ghost went to a screen that said it could not finish and gave me 3 choices I think they were "return to windows" " Try again" or "Curl up and die" I choose "try again" several times and it always craped out around 20%. The D: drive ghosted OK.
I ran the DataLifegaurd tools on the drive and it passed the quick test but failed the extended test but said it repaired the errors it found. After rebooting I got the Raid Error from my Intel software again. Ran the DLGDIAG several time and every time it said it fixed it. But it is still reporting errors. WD is sending another drive. But I would still like to get my C: ghosted. I tried to "Clone" it as well with Norton. No Go. I also tried copying it over to my IDE 80 drive with the WD disk tools. But after rebooting without they array present windows gave me some nonsense about unable to verify liscense or some such. So Itried a repair install of XP on the copied image on the 80 gig drive. It booted and went to the desktop but the install was really screwed up. All drive letters were wrong so no shortcuts worked I renamed the D: drive and that was OK and restored some functionality but the bootdrive was now E: and I can't change it. So there is still a bunch of stuff screwed.
So It looks like I will be ghosting from an image that is about 3 weeks old. Better than a new install. I will lose some emails but that is about it. I can reinstall "Far Cry " (all 5 cd's ) I have already backed up saved games.
What a Pisser !
So the long and short of it is what do I do when I get my new drive. I am pretty fed up with stripped raid.
Any suggestions ?
here is what I have to work with.
2 120 SATA drives and 1 80 IDE drive
onboard Intel controller
Thanks
Scott
Oh yah it is still running just fine. Who now for how long though.
I ran the DataLifegaurd tools on the drive and it passed the quick test but failed the extended test but said it repaired the errors it found. After rebooting I got the Raid Error from my Intel software again. Ran the DLGDIAG several time and every time it said it fixed it. But it is still reporting errors. WD is sending another drive. But I would still like to get my C: ghosted. I tried to "Clone" it as well with Norton. No Go. I also tried copying it over to my IDE 80 drive with the WD disk tools. But after rebooting without they array present windows gave me some nonsense about unable to verify liscense or some such. So Itried a repair install of XP on the copied image on the 80 gig drive. It booted and went to the desktop but the install was really screwed up. All drive letters were wrong so no shortcuts worked I renamed the D: drive and that was OK and restored some functionality but the bootdrive was now E: and I can't change it. So there is still a bunch of stuff screwed.
So It looks like I will be ghosting from an image that is about 3 weeks old. Better than a new install. I will lose some emails but that is about it. I can reinstall "Far Cry " (all 5 cd's ) I have already backed up saved games.
What a Pisser !
So the long and short of it is what do I do when I get my new drive. I am pretty fed up with stripped raid.
Any suggestions ?
here is what I have to work with.
2 120 SATA drives and 1 80 IDE drive
onboard Intel controller
Thanks
Scott
Oh yah it is still running just fine. Who now for how long though.
0
Comments
Tex
Ya, I sure would like to give it a try. I did try to do it in "Disk management" It said it cannot rename a boot drive. And I could not find any options in Partition Magic 8.
But I am all ears ???
Thanks
Scott
You can for instance search for "C:\program" and find 7000 entries and in one swoop change them all to "E:\program" to change the path in the registry to all the programs that don't work from c: to E:
Anyway you get my drift here. The goal is to change the old drive letter to the new one in your registry using a registry editing tool. You can't change the actual OS drive letter but you can change the old one in the reg to your current one and everything will work again. I am sure there are other tools but I use registry crawler. A 30 day free download version is available just google for it or email me.
I have had to do this a gazillion times.
Tex
I will give it a shot later and let you know if it works.
While I have your ear, I have a "mirrored array" question. Lets say I set the 2 Sata drives up in a mirrored array. Ghost my system and partitions back to it. And a few months down the road another one of these drives starts smoking whinning and comes to a screeching halt.
Would my system still run on the one good drive ? Or is the data just "safe" on the good drive until I rebuild the array with a new drive?
Thanks again
Scott
tex
You are the man when it comes to raid.
Csimon
Sorry, I did not see your question before.
Yes my OS is on the array.
Scott
Had I listened to tex at the time and placed my boot sector & OS on the single drive and done a software array I am positive I would still have both drives in array today.
My point is ...booting to the array really wears the hard disks in the array (escpecially the one assigned as the boot disk) much quicker than it does just a single drive especially if you reboot often.
Tex can explain better about the soft array.
Tex
I know a guy that does video processing and bruning to DVD for a business and for him Raid 0 is the right application... I just play around and so I have come to the belief that I really don't need it... I know that I would not raid my OS ever....
My $.02,
"g"
When I raid 6 or 8 really fast scsi drives witha hardware raid controller then its totally differant.
Tex
I have never, never ever, had a corrupt array, even with my old dodgy pair of 75GXP's, I've never lost an array in any shape or form, and I'm pretty rough with them sometimes.
RAID isn't for everybody, granted, but whether it's used by the everyday user, or the video processing guru, RAID is faster than standard single disk setups, by quite a bit if setup properly. Also, the performance advantage is just as obvious when you're copying a basic file as it is to when you're processing a video clip. It is just more of a necessity to the video guru, no more of a benefit.
What can I say, I'm a RAID fanboy. Queen of the Stripe, if you will.
Tex
All 3 drives are in the typical location bottom front. There is an inch between all drives and the bottom of the case. There is a 120mm fan centered on the 3 drives pulling air in from the front of the case and blowing over the drives. I have a temp probe from my fan controler taped to one of the SATA drives. It never gets over 32 C. Even when ripping DVD's. The closest Optical drive is about 6 inches from the top HD ( which is the 80 IDE )
The Controler is Intel 82801ER onboard Sata. The Board is an Intel D875PBZLK. Now I am not sure where the fault lies but I have lost the array twice inside of 8 months both times it was a flakey drive. I do not know if a controler could cause a drive to go south.
Spinner
"Queen" ???
I am having a hard time deciding how to rebuild the array when the new drive gets here. I feel like I am limping along when running on the IDE drive. It is very obvious even when doing routine stuff. I to am a "stripped" fanboy. (but I will not go so far as to say "Queen" ) But the old addage "once bitten twice shy"......How about twice bitten?
Scott
I agree, I run the 3ware 7504 PATA raid card my self. Paid like 350 for it 3 years ago.
I did seperate the IDE raid from the SCSI raid in my post and I agree that if you go true SCSI raid like from an LSI mega raid card or sometihng especially if you can plug it in a 64/66 slot then you've got something and the likely hood that a SCSI drive is going to crap out on you is slim...
However I have already had 2 of WD's enterprise IDE drives crap out on me, one was brand new and the other was the replacement of the afore mentioned new one... Given that they are the only 2 drives that I have ever owned that crapped out on me and I would say that the odds are for most that the likely hood of a drive failure is remote. But that being said I have had 2 of them and even though I have drive image and make backups on a scheduled daily basis to me the pain in the butt that you have to go through to rebuild and restore your array is more than I want to deal with. Not to mention that I don't have an extra drive or 2 just laying around that I can use unitl the crapped drive is replaced.
Example 2-100gb drives raid 0, system backed up and one drive fails... Unless you have a third 100gb drive lying around your system is going to be down a min of 1-2 weeks while you deal with RMA's. To me thats just not worth the effort.
So my Raid 1 I had a drive give out, mirrored no big deal... So I didn't really loose data but I'm still going to have to wait for the RMA process to complete before I can restore my array and then I have to hope that the referbed drive that they will send me will last.. yea right... the one thing I can say about the RMA process is that it sucks that you get a referbushed product that has a half-life of the orgional product that went belly-up...
I don't know about your guy'es systems but I have so much crap configured that I use to teach my self and play with that I really hate to have to go back and reconfigure all of the stuff especially the servers that I run...
For a novice like myself trying to configure Appache and Mysql server along with a copy of Coldfusion and stuff was a pretty big pain in the butt and I'm still not sure how I did it... LOL
I think if I had 5, 10 or 20gb SCSI drives and one of those LSI cards I would not hesitate to run my OS and stuff on it. I would use raid 5 or 10 for some safety but even with the SCSI I probably wouldn't run raid 0...
but thats just me,
"g"
The reason I asked about cooling was a lot of the repeated failures where a single guy is losing multiple drives are also from guys going to great lengths to cool them with add-on hard drive coolers, and big fans, and especially in the new lightweight cases and this creates a problem they never imagained. The add on coolers and big fans teamed with lightweight cases and even when mounted near high speed cdroms allow vibrat5ions from teh fans or cdroms to effect the hard drives. Remember the disks are turning at 7 to 10k and a there is a very thin film of sir the heads are floating on. The older cases were stell and heavy so when you bolted the drive in it was pretty stable. but now withthe flimsy lightweight cases a high speed cdrom can cause enough vibration when it cranks up to make a nearby harddisks heads chatter against the disk surface. As can a nearby high speed fan or the nifty hard disk coolers at times. So it seems at times the guys trying hardest to NOT have probs inadvertantly create more probs for themselves and the extreme lightweight cases of the last few years just exasperate the problem no end.
I wsn't asking because I thought the disks were to hot. But wondering if other fans or drives nearby could be causing the drives to fail.
I have asked a couple guys that were claiming 3 or 4 drives failed in the last 9 months etc... many times with expensive all aluminum fancy butt high dollar cases..... add on fancy harddrive coolers with fans bolted to the bottom and crap.... and I ask them to just reach in and hold their hand on the drives and run their cdroms etc... and they post back just cussing a blue streak as the thing is vibrating like bitch. They are in effect shaking those drives to an early grave.
Not saying thats YOUR problem but something to consider.
Tex
W.D. was nice enough to send me 160Gig drives in exchange due to them being out of stock on the 120's when the first one died and me requesting the same to even things up so I could raid them up again (considering I bought them solely for that) and I haven't had any problems out of the 160's yet.
I've seen other folks in this forum that have had the 120Gig SATA W.D.'s croak on them as well so my first thought about that drive is to stay away from it.
I'd have to say though that I've run striped arrays with more success than failures myself and with great performance gains and I'm not terribly worried about running it on a good set of drives (new, quality) because the performance gains are simply fantastic compared to a single drive.
I hope you get things straightened out and don't let one failure scare you.
Nobody here is a bigger benchmark whore then I am and I would bet no one here has set up more raid arrays in the last ten years, both IDE and scsi as I have. And the performance gains for stuff like runnng your OS in raid-0 is marginal in real life compared against running multiple disks and setting up the pagefile and tempfiles on a seperate disk anyway. IDE raid has clear advantageous for long streaming data files. scsi raid is a differant beast entirely.
We will politely agree to disagree about your "perceived" performance gains.
Tex
A hearty congratulations on running raid for the last ten years, I've just recently gotten into computers compared to most of you so I don't have ten years to draw upon but in the time I've been doing this I have run raid on every system I've had except 3 and one of those was a Shuttle cube with no provisions for an array.
I find that on average my boot times are about the same but when I'm loading apps when I'm in windows the load times are far better and the benchmarks I've run tend to bear this out as in my current config the only drives faster than mine in Sandra are 10k SATA arrays and SCSI.
My average read times are over 80 MPS for the dual 160's and I defrag my disks every other day and run diskeeper on a boot-time defrag twice a month to keep my MFT nice and fragment free.
Disagree with me if you wish, I have tons of free time on my hands and have nothing better to do than tear my rig apart and re-configure it six ways 'til Sunday and I've run it and all my past rigs in every conceivable configuration that was possible in order to gain the last iota of performance out of them.
I may be relatively new to computers (4 yrs now) but I've been a hot rodder all of my life.
Next. Exactly how much time did you spend optimizing the performance of those raptors as single non raided drives? By dividing the I-O across the seperate drives and tweaking the cluster size... moving the pagefile etc...
I bet I can take a good guess at the answer to that one also.
But hey if you want a seperate oppinion fron other experts I highly suggest you pop on over to storagereview.com and post the question in their forums. They specialize in disk subsystems and the performance of disk subsystems. Many of them work professionaly in jobs relating specifically to disk subsystems and its a great place to debunk many of the myths being repeated today.
Go ask them how much of a performance gain you should expect by raiding your two raptors for the type of usage that an average desktop user will perform and ESPECIALLY how much faster the OS is gonna be as opposed to just trying to ballance the I-O manually.
Tex
Anyway...As far as my problem , I did not use your app tex. But resolved the issue another way. The OS was still working on the array, Ghost would not ghost but windows back up would . So I made a fresh Windows backup .bkf. and stored it on a partition on the 80. I then disconnected the array leaving only the 80. I used a 3 weekold Ghost image and ghosted C: to the bougus partition on the 80. It booted up just fine. I then ran windows back up and restored it to current config. Everything worked except Norton AV. I had to reinstall that. But that was the only Glitch. I did not even lose a single email.
Thanks for all the support and help guys ! I will now have a good image to ghost over to the new drives when they arrive. Although I am not sure what they will be yet...1...0...or JBOD.
Please continue with this "Spirited" disscussion, It is helping me decide which way to go.
I was playing "Far Cry" last night. ( off the 80 ) and it takes much longer to load and change maps. I mean a lot longer.
Tex
My case is steel and the 120 fan is low speed and the optical drives are both DVD drives. Neither drive vibrates like the 52X in one of my other machines. But that is a very interesting phenom. As far as secure mounting... I don't like screws, I hold'm in place with really big rare earth magnets. ( just kidding )
Thanks
Scott
I am only stating my own personal experiences, but to say RAID 0 (with 2 disks) is no better performance wise than running a single IDE PATA or SATA drive, is well not true. RAID is high maintenance yes, but certainly not as moot as you are suggesting. No 'one' benchmark I have encountered can appropriately and definitively illustrate real world performance, yes, I agree, but by using a handful of different benchmarking applications, ranging from ATTO to Sandra, you can get a pretty good idea how your RAID array will perform in the desktop arena. At least, I believe I have that annalistic ability. But of course, there is no definitive benchmark, or combination for really anything these days, even for GPU benching. But like I said, I don't often benchmark, I don't need to, I can tell the difference between most computers, ones with and without RAID arrays. Sure, I bet some highly tuned PC's could fool me, but I'm not talking about playing silly little games to pump my ego, I'm talking about typical real world performance.
Even if expertly tweaked and tuned non-RAID 0 systems can bridge the gap between non-RAID 0 and RAID 0 setups, the average Joe isn't the legend of a man you are and probably hasn't the knowledge or ability or even willingness to set up their hard disk partitions in the most performance beneficial manner.
I mean, you yourself taught me a lot of the things I have come to know about RAID, but in my opinion, though I feel what you are saying for the most part is correct, I just can't help but think your interpretation of your own findings is incorrect and exaggerated in an attempt to squash someone else's thoughts.
This started off as a nice little discussion, looking at what setups are best for different user groups, now however, as typical of your behaviour of recent, you've decided to use this as an opportunity to try and demean and patronise other poster's, e.g. myself and perhaps one other.
I have always shown you the highest level of respect here, and back at Icrontic. There is no doubt in my mind you are one of the most knowledgable people in these forums, and I have been pleased you have become quite active of recent. But your manner towards people is starting to irritate me. We've known each other for quite some time, and I would have hoped by now you would show me a least a tiny ounce of respect.
Cheers
p.s (Original topic only now please, if people wish to continue this discussion please feel free to start a new thread)
EDIT - (Here: http://www.short-media.com/forum/showthread.php?p=149028#post149028)