Intel Raid rebuild keeps restarting

[FalconNL]

Hello everyone,

A few days ago Winamp locked up while playing a song. Ok, Ctrl-alt-del and end it. Hm, nothing happens. Kill explorer too. Then I get errors about not being able to save some files I was downloading. Ok, this is starting to sound like a harddisk problem, possibly related to the short power outage I had that morning. Restart, and indeed one of my disks is showing Failed, with the Raid5 array degraded. As expected. Last time I had a problem like this rebuilding the array solved it, so I go back to Windows and start the Intel Matrix Storage manager. One of the drives has a white-cross-in-red-circle icon over it and I can't find the rebuild option. Right-click on the drive in question and click "Restore as normal drive" or something to that effect. Now I can rebuild the array and all will be back to normal... or so I thought. This morning it has restarted rebuilding the array for the second time now (i.e. the third time overall). Obviously something is going wrong with the rebuilding.

Possible reasons:
1 - The hard drive is dead. This seems unlikely, since the rebuilding status reaches at least 98% (I went back to making lunch for a minute and when I looked again it was back at 0%)
2 - There's a bad sector or something on the disk. Does this cause rebuild failures and if so, it is possible to run checkdisk during the re building?
3 - Suggestions welcome.

Details:
Motherboard: Asus P5K Deluxe/WiFi
Raid: Intel Matrix Raid5
Disks: 4x 500GB (raid), 1x 74GB (non-raid)
Activities during rebuilding: browsing, which should only access my non-raid system disk

Open questions:
Is it safe to reboot/power off during a raid rebuild (e.g. if one of the disks turns out to be dead)?
Could I safely make a backup to an external HD while rebuilding?
Is the raid rebuild supposed to take around 20 hours for a 500GB disk? This seems a bit long to me.

Any help on this would be greatly appreciated.
Thanks in advance.

 

[apache_lives]

test your hdd's first up

chkdsk will test the software/file level of the storage, not disk by disk and prolly wont find issues

 

[FalconNL]

Ok, how do I test the disks themselves?

 

[croc]

Your HD MFG's website should have tools available for drive testing. You did not mention drive MFG, or version of Intel raid.

 

[FalconNL]

My harddisk manufacturer is Samsung. I've had a look at their site and they have an ES-tool available. However, this tool must be put on a boot floppy/CD and it recommends to backup the data first.

This leads me back to my previous two questions:
- Can I safely backup data during a rebuild?
- Can I safely reboot during a rebuild?

Thanks again for the help.

 

[croc]

In a raid 5 you 'should' be able to remove the offending drive, plug it in to another sata port and do your tests. The array will take quite a bit of time to recover from that, so expect a lot of actvity...

Once you have tested / replaced yor drive, you can then add it back into your array, and once again the process of re-building the array will take quite a bit of time.

Your mileage may vary, so take this with a grain of salt.

 

[rockchalk]

<snip>
This leads me back to my previous two questions:
- Can I safely backup data during a rebuild?
- Can I safely reboot during a rebuild?
</snip>

I'm not sure about all RAID stacks, but with the Intel Matrix Storage solution, you can access the volume (read/ write) while the rebuild is in progress. And yes, you can safely reboot during a rebuild.

 

[croc]

With a MB based RAID controller? I'd not trust the controller enough for either operation during the re-build.

 

[cjl]

Intel raid works fine with a degraded array, though if you're rebuilding while using, it will run quite slowly.

 

[croc]

Intel raid works fine with a degraded array, though if you're rebuilding while using, it will run quite slowly.

Not arguing here, but the questions were about backing up and / or restarting the array while the restore process was happening. Given that even Intel's controllers tend to off-load the XOR functionality to the CPU, plus its tendency to rely on drive cache... I'd be hesitant to get a reliable backup during the rebuild process, and would be VERY nervous about a re-boot.

Maybe one day when I'm bored I'll hook up a few spare drives and test those theories...