Re: AIX ASM and raw disk...big hassle....need help

From: oracle_man <oracle_man_at_yahoo.com>
Date: 29 Nov 2006 10:00:36 -0800
Message-ID: <1164823236.318163.64670@80g2000cwy.googlegroups.com>

hpuxrac wrote:
> oracle_man wrote:
> > hpuxrac wrote:
> > > oracle_man wrote:
> > > > All,
> > > >
> > > > I have 2 AIX 5.3 clustered systems, oracle 10.2 STANDARD Edition, and a
> > > > few multipathed(SDDPCM) raw disks. I am unable to find any notes
> > > > regarding my failure type. Below are the commands used to create my
> > > > disk group, and the device files listing.
> > > >
> > > > I did the mknod myself for the device files, as they are supposed to be
> > > > the same name on each node.
> > > >
> > > > Any insight is appreciate.
> > > >
> > > > SQL> /
> > > > CREATE DISKGROUP DG_01 External REDUNDANCY DISK '/dev/ora_raw_1' SIZE
> > > > 51200M
> > > > *
> > > > ERROR at line 1:
> > > > ORA-15018: diskgroup cannot be created
> > > > IBM AIX RISC System/6000 Error: 22: Invalid argument
> > > > Additional information: -1
> > > > Additional information: 655360
> > > > IBM AIX RISC System/6000 Error: 22: Invalid argument
> > > > Additional information: -1
> > > > Additional information: 655360
> > > > IBM AIX RISC System/6000 Error: 22: Invalid argument
> > > > Additional information: -1
> > > > Additional information: 655360
> > > > IBM AIX RISC System/6000 Error: 22: Invalid argument
> > > > Additional information: -1
> > > > Additional information: 655360
> > > > IBM AIX RISC System/6000 Error: 22: Invalid argument
> > > > Additional information: -1
> > > > Additional information: 655360
> > > > IBM AIX RISC System/6000 Error: 22: Invalid argument
> > > > Additional information: -1
> > > > Additional information: 655360
> > > > IBM AIX RISC System/6000 Error: 22: Invalid argument
> > > > Additional information: -1
> > > > Additional information: 655360
> > > > IBM AIX RISC System/6000 Error: 22: Invalid argument
> > > > Additional information: -1
> > > > Additional information: 655360
> > > > IBM AIX RISC System/6000 Error: 22: Invalid argument
> > > > Additional information: -1
> > > > Additional infor
> > > >
> > > > Here are the raw disk files used for the
> > > > group:
> > > >
> > > > +ASM2:oracle_at_mango1:/u01/app/oracle/product/10.2.0/dbs $ ll /dev/ora_*
> > > > crwxrwxrwx 1 oracle dba 41, 0 Nov 17 17:55 /dev/ora_raw_2
> > > > crwxrwxrwx 1 oracle dba 41, 2 Nov 17 17:56 /dev/ora_raw_3
> > > > crwxrwxrwx 1 oracle dba 41, 1 Nov 17 17:57 /dev/ora_raw_4
> > > > crwxrwxrwx 1 oracle dba 41, 3 Nov 17 18:05 /dev/ora_raw_1
> > > > +ASM2:oracle_at_mango1:/u01/app/oracle/product/10.2.0/dbs $
> > >
> > > Not running AIX at my shop but I guess the first question I would ask
> > > you is did you go through all the steps ( whatever they are on AIX ) to
> > > properly create a raw volume?
> > >
> > > On hpux that would be vgcreate and lvcreate then change permissions on
> > > the raw partitions.
> > >
> > > Looks to me as if this is an item that you should probably work with
> > > oracle support on.
> >
> > I have an SR open with support at sev2. They haven't touched it in 5
> > days. As for vgcreate, et....My unix admins handled that aspect, and I
> > assume they did the right thing, although I am somewhat suspect of
> > their actions. The AIX install guide has a number of hoops to jump
> > though(because AIX sucks). One of the steps is a mknod on each node if
> > the device file names are not the same.

>

> I left out some stuff such as mknod etc because they are part of the
> process ( at least in hpux ) to do a vgcreate.

>

> You can call into oracle support and ask to speak to the on duty
> manager that's responsible for your sr. Once they call back ( supposed
> to be in half an hour ) then you can try to get them to expedite it.

>

> In the mean time I would look at all the steps that "should have been"
> done to setup the raw disk partitions.

>

> Can you use dd command to look at the first 10 blocks? If so do they
> look like what you expect?

>

> Oracle support has metalink notes on raw device setup and configuration
> for the various unix platforms and you might want to look at some of
> those notes.

Fixed. It turns out that you have to turn off FastPath via smit aio. Otherwize, my time was wasted for days.

rich Received on Wed Nov 29 2006 - 12:00:36 CST