unable to create 2 advfs domains on raid device

[Sarika]

I have 1.6 TB of raid array & i want to create 2 advfs doamins on that.
I have created 2 partitions of 800 GB each.
These devices are dsk5a & dsk5g
I am able to create advfs domain for dsk5a( 1st partition ).
But while creating 2nd advfs domain it gives following error:-
mkfdmn: domain init error. Read only file system.
mkfdmn: can't create new domain

What could be the reason ?
Please help.

[Han Pilmeyer]

Which version?
Is it a cluster?

Can you show us the partition table of the disk (disklabel) in combination with the "hwmgr -get attr" information for the disk?

Also make sure that there are no errors in the error log.

I suspect a problem in the disklabel.

[Hein van den Heuvel]

Hmm, I only recall seeing an error like this when running in standamone mode and the /etc/fdmns structure not being writeable (fixed with mount -u /). But here one domain was already created, so that rules that out.

Maybe the device is truly read-only?
What is the raid controller used.
What (full) status does it show for the second unit? Is that different from the first?

Is the disklabel correct?
Right number of blocks?
Right device type?
Compare disklabel -r for both units
Are they the same?
Should they be? (minor size difference?)

hth,
Hein.

[Ivan Ferreira]

Ok, you have a RAID device (One LUN) and created two partitions.

You can create the domain on the first partition, but you can't in the second.

Can you create the domain using the dsk5c device?

Why do you want to have separate partitions (domains) when you can have one domain with multiple filesets?

[Hein van den Heuvel]

Ooops, I read too fast, and was thinking you presented 2 devices to the system, keeping each under 1TB.

Ivan correctly points out that you describe a single, very large, partitioned disk. So most of my questions make no sense, but you still need to doublecheck disklabel output.

By partitioning you 'garantuee' maximum head movement between the domains, where filesets may allows them to fill up in sync. Mind you, this may be a moot point if the raid system is allowed to dynamically allocate actual storage as needed and not coerced with an initial 'all block write' write.

I'll stop rambling now,
Hein.

[Michael Schulte zur Sur]

Hi,

what kind of storage system are you using?

greetings,

Michael