A month back, there was discussion about hotswapping IDE disks.
I have a solution that seems to work, indirectly, using the SanMax
InClose PMD96-USB2 Mobile Dock. This cage is physically compatable
with the PMD96i direct IDE cages, but has a translator from IDE
(LBA48!) to USB2, which I connect to a USB2 PCI card in my system.
With USB hotswap enabled, the system seems to recognize when
disks are added and removed. This allows me to do backups to
cheap big IDE hard drives, and swap them out for safekeeping.
For more information, look at my web page in progress:
http://www.keithl.com/linuxbackup.html
I am still working on this, so your mileage may vary. I am
incompetent at software, so a more clueful person is welcome
to take over and do this right. In any case, this provides a
functional solution for IDE hotswap without kernel patches, and
that might free up some wizard time for other needed tasks.
Keith
--
Keith Lofstrom [email protected] Voice (503)-520-1993
KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon"
Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
> A month back, there was discussion about hotswapping IDE disks.
> I have a solution that seems to work, indirectly, using the SanMax
> InClose PMD96-USB2 Mobile Dock. This cage is physically compatable
> with the PMD96i direct IDE cages, but has a translator from IDE
> (LBA48!) to USB2, which I connect to a USB2 PCI card in my system.
> With USB hotswap enabled, the system seems to recognize when
> disks are added and removed. This allows me to do backups to
> cheap big IDE hard drives, and swap them out for safekeeping.
> For more information, look at my web page in progress:
>
> http://www.keithl.com/linuxbackup.html
>
> I am still working on this, so your mileage may vary. I am
> incompetent at software, so a more clueful person is welcome
> to take over and do this right. In any case, this provides a
> functional solution for IDE hotswap without kernel patches, and
> that might free up some wizard time for other needed tasks.
>
> Keith
This isn't IDE hotswap, its USB hotswap with USB Mass-Storage devices.
Mark.
- --
Mark Watts
Senior Systems Engineer
QinetiQ TIM
St Andrews Road, Malvern
GPG Public Key ID: 455420ED
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)
iD8DBQE/XestBn4EFUVUIO0RAkUDAJwKZFVCVRMgfiqAvHodFKHkV+fCfACeOCMO
/rAEWJkRoogVkb+k3ebZ92Y=
=teqy
-----END PGP SIGNATURE-----
On Tue, Sep 09, 2003 at 04:01:01PM +0100, Mark Watts wrote:
> This isn't IDE hotswap, its USB hotswap with USB Mass-Storage devices.
Taxinomically, from a kernel-centric viewpoint, you are correct.
However, from the point of view of users (remember them?) it is
an LBA48 ATA6 IDE device in a bare removable cage that is getting
loaded and unloaded, while the system is running. The fact that,
inside the server case, there are two extra interfaces and an
always-connected USB2 cable is really not that important. That
is why I called it "functional" IDE hotswap.
True, the bandwidth is lower, it is a distasteful kludge, and
it is more expensive than a hypothetical, purely IDE solution.
However, a pure solution is not now available, and the pursuit
of that may be a distraction from other pressing issues.
My real concern with the method is that somebody may change
chipsets for the USB2 to IDE interface, and the hotswap will no
longer work with the new chipset. Or the USB mass storage module
might get "improved". Thus a direct IDE-chipset-driven hotswap
scheme is still desirable, but perhaps not with the same urgency.
Keith
--
Keith Lofstrom [email protected] Voice (503)-520-1993
KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon"
Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs
On Maw, 2003-09-09 at 17:33, Keith Lofstrom wrote:
> it is more expensive than a hypothetical, purely IDE solution.
> However, a pure solution is not now available, and the pursuit
> of that may be a distraction from other pressing issues.
The IDE solution isnt hypothetical any more, its in 2.4.23-pre
at least for some chipsets and adding more isnt too hard