Windows 2000 Cluster & RA4100 Raid Array

[Jon Daniels_1]

Hi,

We are experiencing a problem of slow performance on a couple of the logical drives on a Windows 2000 cluster.

When running the array configuration utility we get the following error message ...

Background parity initialization is currently queued or in progress on Compaq RA4000/4100 Controller in RAID Array 8008DBX10067 on the following logical drives:

Logical Drive 2
Logical Drive 3

This is a normal operation that is necessary to initialize logical drives that have a fault tolerance with parity. If background parity initialization is queued, it will start when I/O is performed on the drives. When background parity initialization completes, the performance of the logical drives will improve.

Which lead us to the following technical article ...

http://www3.compaq.com/support/reference_library/viewdocument.asp?countrycode=1000&prodid=499&source=EC030424_CW01.xml&dt=3&docid=20228

We have two questions on the above ...

1. Does anyone know how long this should take.

2. This may be more a MS question but where the article says to remove all logical drives on which parity initialisation is not complete from the cluster resource whould taking them offline line in the cluster be sufficient ... or do we need to go further ... would this lose any data that is on the logical device?

Thanks

[Dave_180]

Generally it is something that is done any time a new array is created. I haven't seen the message ever last more than a day.

If it is going longer than that - I would recommend updating the controller firmware - should be version 2.60.

Also, look at updating the drive firmware and add the M&P patch.

If it were me - I'd leave one of the cluster nodes up and let that server handle the process - of course with the cluster service stopped.

[Jeff Allen_5]

That one storage box does have this oddity that cause this to happen if Disk I/O is too heavy before the process was able to complete. The idea is that you have got to cease the writing to the logical volumes so it can finish. It should take only hours, not days to complete. So, yes, you can just take the cluster offline and let the process continue.

[Iain Hamilton_4]

These resolutions are interesting but I thought the point of a cluster was that it was never shut down. I have replaced 18gig drives with 72gig in my cluster and the said error message has been there for weeks. As I work in a hospital we cant shut it down completely for any length of time, so how do I start the initialization. If I shut down for 15mins will that start it and if I do this at various times will it pick up where it left off and continue the process??