IBM X Series RAID - Rebuilding Single Drive in RAID5. 1

[Raedeke]

We have a 3-4 Year old xSeries 226 server.
Recently we had a drive go down and show defunct in the Raid Manager. We bought a replacement drive from third party (not IBM - they wanted $600 and looked to take 6 weeks to acquire) The drive is identical - 73.5G U320, etc. (same model number) We loaded it in and told it to rebuild... that was about 12 hours ago. In the log it has an entry indicating that it knows it's not a IBM drive, could this be keeping it from rebuilding? The amber light on the face plate is blinking, but there is no progress or monitoring available to show me that it's actually doing anything. Does anyone have experience on using third party drives - is there a way to check progress - does 12+ hours for a rebuild on a 73G drive sound correct? Thanks in advance for the assistance.

 

[ibmtech65]

Raid manager should show the progress.

If you have a blinking amber light and all the other drives have pretty much solid green it is rebuilding.

Also are you sure it is a straight raid 5 or is it something like a raid 5E or 5EE?

 

[Raedeke]

Interesting you ask about the RAID5 - In the properties of the logical drive it does show it as a RAID5EE in an EXPANDED state. It also says that it is not protected by hot spare and the write cache mode is Enabled(write-back)
Not sure I know what RAID5EE is...

Regarding the lights -
The drive it says is rebuilding is blinking amber -
No other lights flashing

Sounds like the 5EE is what has got us confused...

 

[Raedeke]

OK - so maybe I need to ask a question.
Is there any special instructions at to how to rebuild a array that is configured in RAID 5EE?

At this point the thing could say it's rebuilding for the next year and I can't believe it would have done anything. It's been working for over 24 hours now and I've got to believe we did something wrong.
Can I stop the rebuilding by making the drive defunct and starting over?

 

[ibmtech65]

5EE is going to take a long time.

That is just the nature of it. When a drive fails it then compacts to a standard raid 5.

Once you replace the drive it then has to expand and rebuild. I have seen that entire process, compact, expand and rebuild take 3 days.

That is the trade off of raid 5EE versus standard 5. The 5EE is faster performance but you killed on the other end.

 

[Raedeke]

I appreciate the information on how it's supposed to happen.
My problem is I don't trust that it's actually working.
When it goes into rebuilding mode - I get no progress bar.
Although the new drive is flashing amber, there is no activity on the other drives.

Is there some prep work I need to do on the new drive to ensure that the controller recognizes it as a valid drive and it's set up correctly.

The new drives I have bought are not from IBM, as they wanted 6 times as much and indicated a 6 week lead time on their site - they are however the same manufacturer, size, speed, etc. In fact they are the same model number - just no IBM label.... does this all make sense -
should I just run the rebuild and let it go for a few days?

 

[ibmtech65]

There should be drive activity on all drives.

I would right click on the rebuilding drive and mark it defunct and then see if the array will compact. Sometimes what happens is the array has not finished compacting down to a standard raid 5 and when the replacement drive is added in it throws it off.

Could also be the none standard IBM hard drive.

 

[Raedeke]

Any idea where I find where it might tell me if the array has been compacted yet?

Unfortunatly, when the drive when down, it seemed that there may have been another bad drive in the array. I came back up and we had access to info all day yesterday.
Today it looks like active directory is corruct and has to be restored, etc. so the server is down at this point awaiting the arrival of my tech. so at this point I can only work the issue from memory -

So let me just ask one more question about procedure.
When the original drive when defunct, I should have been able to pull it out, load in a new one and put it back - and have the raid rebuild automatically? My experience was that it read the new drive as defunct immediatly and I had to tell it to rebuild... at which point the amber light started to flash - got the rebuild indicator on the drive, but no action on other three. How long could the compacting have taken? - it's only 3 - 73.5G drives?

 

[Raedeke]

New Information.
Over the weekend additional drives have "gone bad".
The drive at ID 0 on our mirrored set all of a sudden shows it's defuct.
At one point so did ID 1 - which meant we booted from the Raid boot disk - but was able to put it back on line.
Also, another drive in the RAID5 showed up as defunct and it too allowed me to make it on-line - the RAID Manager showed that the drive I brought back up did have some PFA errors, but showed none for the drives that failed and needed to be rebuilt. I'm of the belief that the cable, backplane or controller are actually bad and not the drives, but then I can't really tell. Once we have the data off- I can really got to town on diagnosis. But your thoughts are appreciated and welcome at any time.
Thanks

 

[ibmtech65]

If you have a PFA you better plan on getting the drive replaced soon. I would also recommend you update the hard drive firmware.

http://www-304.ibm.com/jct01004c/sy...cdisplay?brandind=5000008&lndocid=MIGR-4GSU4F

Create the CD with the .ISO file then boot into ctrl-i and in advance functions restore to defaults. This sets all your hard drives to Ready. Then boot to the CD and run the update.

Then you can go back into ctrl-i and in advance functions copy the configuration from drives back to the controller which will restore your config.

They say that the hard drive update program will update defunct drives but it is hit or miss so best thing is to make then all Ready then do the update.

 

[Raedeke]

Thanks for the information.
Once I have exchange and data moved over to new server, I will be more aggressive and try your suggestions. Right now I'm just trying to get the data off as quickly as I can.

The server is a good box and I hope to be able to rebuild it and keep it as a spare -

thanks again for the suggestions and the help.
R