slow rebuild with HP 642 array adapter

[Berry Dijkstra]

Hi,

we have a ML370 win2000 server with a HP642 array adapter 6 x 72 GB disks, one of wich was an on-line spare. over a very short period of time 2 disks failed. we replaced the 2 defect disks.

We did this almost 3 days ago, and the controller is still rebuilding the array.
rebuild priority is set to high, expansion is still set to low. (should this also be high ?)

the ACU display shows a rebuild percentage up to 40% and then it jumps back to 0% over and over again. is this normal ?

in the manuals i noticed that the rebuild process takes about 15 minutes per GB, is this correct with ultra 320 10K disks ?

a would apreciate som reassuring remarks :-)

thanks
berry

[JohnWRuffo]

Berry:

The 15 minute per GB is a true statement (without the BBWCE) for the expansion process; espacially if you are expanding online and especially if you have the status bar up all the time.

The 15min/GB rule does not refer to the rebuild process. This should not take so long. I have seen 72GB disks rebuild in 30 to 45 minutes.

When you say it lost two drives, do you mean that it lost one drive, rebuilt on the hot spare and then another drive died? If this is not true you may not have a valid array depending on the Fault Tollerance you selected.

FYI: The rebuild process can only do one drive at a time. So it will have to do the process twice to get back to the original state.

[Berry Dijkstra]

Hi

we noticed that there were 2 disks crashed, in the event log we saw afterwards that the first rebuild on the of-line spare disk took 1 day.
then a few days later the second disk crashed, so i think the array should be in a correct state.

i replaced the first disk (disk 3), and about 1 hour later the system demanded also the replacement of the second disk (disk 0), and after I placed the second disk, i also set the rebuild priority to high. all the disks are on line now, all the orange lights and green lights are flashing continoues.

but what wonders me most is the display that runs from 0% to 40% and then back to 0

thanks for your response, luckily we do have a recent good tape backup :-(

berry

[Greg Carlson]

Berry,

Install and run the online ADU v 2.40.6.0. Post the report.zip here for us to look over.

http://h18007.www1.hp.com/support/files/server/us/download/20061.html

Ciao,
Greg

[Berry Dijkstra]

I ran the utility and to my horror in the first part i saw a remark concerning drive 5.

since I am at this momont at home i was not able to fully read the report. but anyhow it is attached
and later this daty i will fully read this report.

if this drive should be replaced i think this is not possible right now, since de controller is still rebuilding, or am i wrong ?

berry

[Greg Carlson]

Berry,

ID 5 needs to be replaced. It has a very high rate of bad blocks( Re-mapped 4,294,967,294), over 2000 read recovers and 2 predicted failures. It is still spitting out hundreds of errors just 10 minutes prior to your running that report.

The other hdds look like there error counts are more within normal paramaters.
Replace ID 5 ASAP

Ciao,
Greg

[Berry Dijkstra]

Greg,

Thanks, we also noticed it, but we also will have the controller replaced. we seriously think that the disks are just the messengers of a failed controller. Not only disk 5 is giving problems, but disk0 wich was replaced last week also turns up in the event log with bad blocks.

I cannot believe that todays disks are that bad, i also will aks my supplier if it is possible that the powersupply to the disks can cause such malfunctioning

I will post a message if everything is in good working order, and if we can prove cause, i will mention that to

sorry for possible bad language, im from holland :-)

berry

[Berry Dijkstra]

Hi again.

we had the array controller replaced 5 hours ago, and the new controller continued with rebuilding the array, as far as i can see there is no positive development in the rebuilding process.

Is there anywhere in the software a place where i can see at what stage of the rebuilding process teh array is ?

the Acu util goes from 0 to 40 and back again. so this looks useless to me.

berry

[Greg Carlson]

Berry,

Have you already replaced the Array controller and the hdds? If you have replaced all of those and are still not completing rebuilds, it is possible that your array parity is corrupt and you may need to delete your array, recreate it and restore from backup. Your experiences are not normal.

Ciao,
Greg

[Berry Dijkstra]

No we just had the array controller replaced, not the disks, and after that concluded that nothing changes, and now have decided to completely wipe the array and start from scratch.

the vendor/replacer suggested we should still wait... and see,,,,

thanks cooperating and thinking and greetings to you all :-)

berry