Rebuild process RAID5

[BakxM]

Hi all,

Hardware: MSA500, 5x 300GB U320 HDUs
Can anyone explain (or point towards documentation), how the rebuild process in a RAID5 array works when a HDU is replaced?
For what I can find, rebuilding takes about 15min/GB. Watching the rebuild progress-indicator, that's about correct. But when the indicator reached 58% in our particular situation, it jumped towards 100% and finished, leaving me curious as-to-why the sudden 'rush'. Are there factors that speed up this process etc?
Also, do you need to reboot the SA-controller when changing the rebuild-priority and does rebooting or even turning-off the MSA affect the rebuild process?

Thanks for any info!!

Marcel

[Steven Clementi]

Marcel:

Once the drive is replaced, the rebuild process is normally automatic. It starts and runs at the stated rate, or about there and there are several factors that will determine the actual rebuild rate.

Priority setting and actual i/o and/or usage of the array will determine how fast or how slow the drive will rebuild.

You should not have to re-boot when you change the priority settings. It should take effect almost immediately.

You do not want to reboot the MSA or turn it off while it is rebuilding.

Steven

[NMory]

You can check these Documents:

http://h20000.www2.hp.com/bc/docs/support/SupportManual/c00419664/c00419664.pdf

http://h20000.www2.hp.com/bc/docs/support/SupportManual/c00419664/c00419664.pdf

ftp://ftp.compaq.com/pub/products/servers/proliantstorage/drives-enclosures/hp%20hard%20drive%20replacement.pdf

Regards,

LN

[BakxM]

Guys,

Thanks for the info...

Still the question why the leap from 58 towards 100%. Are sectors on a HDU that are 'marked' when not used by i.e. NTFS-Filesystem, so they are skipped in a rebuild process?

What happens when a server IS rebooted, still then there is 1 HDU failing in the Array, does the rebuild process start-over?

Are there performance-monitoring tools for an ArrayController. i.e. sa6400 or MSA500 Cluster Controller?

cheers,
Marcel

[NMory · Accepted solution]

Taking back your initial questions:

a) The sudden jump from 58% to 100% could mean a lot of things:

1-could be that your hard drive has a lot of free space and that's why it jumped to 100% suddenly, because there was nothing more to rebuild.
2-It could be also that something was hanging the utility, a process, of some sort, that wasn't letting your ACU to refresh the progress bar.
3-Maybe the server was too busy attending other tasks, that at some point it just freed some of your resources and that's why the sudden leap to 100%.
4-The controller was busy attending other I/O requests and at somepoint the I/O requests lowered down.
5-Or actually the rebuild process wasn't successful, probably because the hard drive is not that good.

Just make sure you have acces to the rebuilt information, if that is the case then you should be fine, if by some reason you start seeing that you are having troubles accesing this new disk then probably the disk is defective. You can run Array Diagnostics Utility too monitor this HD, just scroll thru the txt document until you see the SCSI ID corresponding to this disk, and check for Hard Reads and Writes, Hardware Errors, Predictive Failures, Retries, etc. of any kind.

b-) There is no need to reboot the Smart Array Controller when you change the priority.

c-)Rebooting or even turning off the MSA, when is doing a Rebuilding is not a good idea, Hardware could be damaged that way. Don't remember really well what happened but if I am not wrong it will start over the whole rebuilding, because according to the Smart Array the rebuilding process wasn't finished, and to avoid a defective rebuild the Smart Array should start it just to maintain the data consistency.

About the performance monitoring tools for the Array Controllers, I think there is not too much of that, only Array Daiagnostics Utility and Array Configuration Utility, or System Management Homepage or SIM.

I don't know if Performance Management Pack will give information about SA Controllers, maybe!!.

Regards,

LN

[BakxM]

Thanks for the info!

[BakxM]

close thread