MD1000 - PERC5e problem

Hi Adrian,

Can you try a different PERC controller or cable?

I'm waiting for a replacement PERC6e. I'm sure the calbes are OK - unlike the old SCSI cables, where a single bad wire could drop a disk or two, the MD cables are serial so they should either work or not?

I have a spare PERC H400 but am reulcftant to try a different PERC family in case this risks my data.

I have never seen such an issue before where I cannot get into0 the config menu, despite having used PERCs since 2000.

At a loss....

I wouldn't advice you to do rebuild in a such dengerouse situation by a couple of reasons:

1) You have only 1st level redundency (raid5) on each MD1000.

2) System shouldn't crash in case of one dead drive, it should work in degrade, so you have some additional issue, may be bad sectors on some of remaining drives. To force rebuild you should choose proper disks with one missing replaced (new), in case if you make mistake replacing wrong drive, after rebuild data-blocks mess garanteed. Rebuild process update all parity blocks on every single disk, and wrong selection will cause to inreversable loss.

I would probably can help you remotely if you be able to connect your both  MD1000 enclosures to HBA controller (None raid), this way we can research logs on back of each individual drives and make virtual reconstrcution of whole array in order to transfer data safe to a separate storage.

Hi Dr.Kiev - thanks for your reply.

I welcome your advice! Pehaps you misunderstood my description of the configuration - the server is running two MD1000's each as a separate volume. The one MD1000 is running fine - it is the second MD1000 (which runs on a separate controller) that has a problem.

The MD1000 with the problem is configured in RAID50 - using both channels of the PERC5e controller and enclosure, so that drives 0-6 are on on channel and 7-14 on the other, giving what I percive as 2 RAID5 arrays put together in RAID50.

I now have a replacement PERC6e controller, which I haven't installed and tried yet. With the current PERC5e controlelr, I can't get beyond POST - I cannot enter the PERC configuration menu to see what it says about the drives/array. neither can I move on to Windows to use the Dell OMSA utility to examine the array.

If you have some ideas of how to proceed, I would welcome them!

Thanks/Adrian

Hi again Dr. Kiev - I hope you understand from my previous reply that the two MD1000's are not striped together - the one MD1000 works fine, the other has this fault.

I have now received a new PERC6e and was going to try this on the faulty MD1000 - but maybe I should wait for your reply before risking data loss?

Best regards / Adrian

If you have  a chance to connect those drives to any HBA controller and image them (byte-to-byte) to a plain images files, that would be great and you will be safe after attept of replacing Perc controller in case if something go wrong. Perc5e and Perc6E is a different generation, i am not sure they are interchangable . The worse thing which can happen is automatical rebuild with wrong configuation (wrong replaced hdd or order of drive, stripe size etc) . Configuration is stored on back of each hdd, there is indetificators, order, type of parity, stripe size etc.

Raid50 is a two raid5 striped each other as raid0 . It is quite complex constrcution. 

Hi Dr. Kiev - thanks for your mail. I don't think I have any suitable HBA with which I can do this, and I have never tried this before so I don't know where to begin!

I haven't tried a new controller yet (because of your warning) but with the existing controller and the original (2 x faulty) drives, I can get into the configuration utility, but it reports two drives as missing, and the menu for Physical disks is not available.

If I try to replace the faulty drives, the drives spin up and POST gives me an alarm, with the Press C option, which doesn't work. Sometimes it proceeds to start the OS, but only as far as the "start windows normally or repair option, which doesn't allow any keyboard input (it seems at this point that the keyboard is dead) so I have to shut down.

The PERC5 and PERC6 are different generations but they use the same driver, so the difference might not be that great. I have migrated between these before without difficulty.

I could really use a chat with anyone from Dell or an experienced user who knows MD1000/PERC5/6e products but I have no idea how to come into contact with anyone. I can be contacted via 

or here.

If those two drives already marked in metadata config as failed, any of existing controller won't let you hook up previuse correct config and won't give you access to volumes as result. Your best option is to find HBA Lsi controller and folow DR instructions how to create virtual array in order to save files. Each raid5 can work in degrade mode only with one failed disk   HBA controllers is relatively cheap (30-50usd) and easy to find.

Thanks again, but I don't understand the method you suggest. As these drives are SATAu I could connect them to a SATA port on another workstation - one at a time, but I don't understand how to saccess them under Windows without initialising them (which is not a good idea) or is there another way to access the data? And if I copy the image for each of the working drives, how do I put it back on a new array? I'm sorry but I just don't understand the method

I do suspect the PERC5e controller because it seems to prevent me entering the config menu or booting the server to OS - so perhaps I have to change the controller anyway?

If drives are sata, even better, if you can plug them in all together to one Pc, that is good. Windows doesn't initialize them by default, it only asks you about. Cancel initializing and put offline all disks , thats enough to work with them with specific data recovery aplication. PM me if you need assistance.