[ale] Sata disk mode query.

[Jack Schneider]

On Thu, 20 Jan 2011 11:44:41 -0500
"Michael B. Trausch" <mike at trausch.us> wrote:
Hi,  All
'Got enough I think, will let you know in 10-12 days when the Release
comes.  I have understood that mdadm metadata has been shifted around
significantly and I have multi-raid arrays from different versions. So
Back-up is in progress for some/most data.  The O/S stuff will go
bye-bye then.

Thanks to all.
Jack  


> > Hi, Greg
> > M/B is a X7DAL-E  Supermicro with a Intel Chipset. I don't have
> > access to the HWinfo file right now..  So can't get the disk drive
> > chips set #,
> > Thanks for the reply!
> > Jack
> >
> > On Thu, 20 Jan 2011 10:31:41 -0500
> > Greg Freemyer <greg.freemyer at gmail.com> wrote:
> >
> >> If your drives are currently /dev/hda, etc. they will change
> >> to /dev/sda, etc.
> >>
> >> Some people really don't like that because they have configs setup
> >> that depend on the name.
> >>
> >> Also /dev/sdX naming is dynamic.  So if you are the type to plug in
> >> drives, etc. from time to time, you need to make sure you don't
> >> depend on your drives being any specific /dev/sdX name.  You can
> >> use mount-by-label etc. to do that, but I admit to finding it more
> >> trouble than its worth and still use /dev/sdX.  I know enough to
> >> make it work in general.
> >>
> >> The good news is that both mdraid and LVM tag the disks/partitions
> >> with identifiers, so they don't depend on the /dev/sdX name.
> >>
> >> Also, AHCI is effectively a controller spec.
> >>
> >> Some implementations are great, others not so great.
> >>
> >> What chipset is it?  Once you know, google for issues.
> >>
> >> == note
> >> If the MB is from the last 2 or so years, I'd expect good support.
> >>
> >> Some of the older stuff is more problematic.
> >>
> >> Greg
> >>
> >> On Thu, Jan 20, 2011 at 10:13 AM, Jack Schneider
> >> <puck at dp-indexing.com> wrote:
> >> > Hi, Alers
> >> > With the pending release of Debian Squeeze I am considering a
> >> > complete O/S reinstall, tons of cruft. My question is: Should I
> >> > change my MoBo disk setup from Pata to Sata AHCI?   I have seen
> >> > both pros and cons on google and Wikipedia.  I have 4 Baracuda
> >> > 7200 rpm sata drives, raid1 and LVM2 used to run VM's. I know
> >> > there are disk savvy folks here with useful opinions and
> >> > experiences... TIA
> >> >
> >> > Long Lurker
> >> > Jack
> >> > _______________________________________________
> >> > Ale mailing list
> >> > Ale at ale.org
> >> > http://mail.ale.org/mailman/listinfo/ale
> >> > See JOBS, ANNOUNCE and SCHOOLS lists at
> >> > http://mail.ale.org/mailman/listinfo
> >> >
> >>
> >>
> >>
> >
> > _______________________________________________
> > Ale mailing list
> > Ale at ale.org
> > http://mail.ale.org/mailman/listinfo/aleificantly 


> On Thu, 2011-01-20 at 11:37 -0500, Randy Ramsdell wrote:
> > Note that the drive sequence letters or what ever can be hard coded
> > in udev. So the disk switching etc... is not an issue. Also note the
> > using UUID is unique and therefor can be used to mount. This
> > becomes particularly important when dealing with iscsi volumes. 
> 
> But do remember that:
> 
>  * Drive names will not stay static if the physical drives are
>    swapped out.  IIRC, udev uses serial numbers in order to ensure
>    that the names are renamed to the expected device node names
>    prior to the majority of userspace opening the drive.
> 
>  * UUIDs will only work if you do not use dd images or LVM snapshots
>    of the volumes housing the filesystems.  The upshot is that in the
>    case of LVM, you get a different way to uniquely address a given
>    filesystem: /dev/${VOLUME_GROUP}/${LOGICAL_VOLUME} will work, and
>    will be unique since you cannot have duplicate volume groups or
>    logical volumes.
> 
>  * Things outside of udev's control (for example, nbd if it is used)
>    will have to have their setup order controlled by something in
>    userspace.
> 
> 	--- Mike