Topic: Controller IDE Driver disk

[jeff at jeffandsteph.com]

Sorry for the delay.

Here's where I stand...

I found that my cpio compression parameters for my 5.6 driver disk were incorrect. I have since recompressed the archive and now the install finds the hpt374 driver (yea...) HOWEVER, it is looking in a directory called 2.4.18-10 (not -5).  Putting the -5 compiled driver in the -10 modules directory leaves me at my current state - the driver can not load because of an unresolved symbol scsi_register_R57c62879.

I suspect this may be due to the kernel upgrade between -5 and -10, so I have found the source code for -10, and will try recompiling with this version of the source.  I'll let you know how I make out.

P.S. My local machine has been off the net while I worked with the 404 driver disk.  I'll reconnect it and send the Athlon driver to my website as promised earlier.

At your service,

Jeff

[Jeff]

I have uploaded the Athlon version of the 404 driver compiled to the 2.5.18-5 source code.  It is located at:
http://www.jeffandsteph.com/athlon/hpt374.o

[Charlie Brady]

Jeff wrote:
 
> I have uploaded the Athlon version of the 404 driver compiled
> to the 2.5.18-5 source code.  It is located at:
> http://www.jeffandsteph.com/athlon/hpt374.o

Do you have the source code there as well? If so, where?

I'm interested to know why you you've gone to so much trouble to build this module? Do you really need the raw speed? Many servers have their disks essentially idle nearly all the time, and the disks can saturate the network over an ATA33 interface.

Charlie

[Seth]

Charlie:
>I'm interested to know why you you've gone to so much trouble to build this module? Do you really need the raw speed? Many servers have their disks essentially idle nearly all the time, and the disks can saturate the network over an ATA33 interface.

Uh, well in my case at least I'm after RAID 1 reliability, not raw speed. If I'd realised there was a problem with hpt374 I would not have bought an older raid controller! .Having got it, it seems a shame not to use it, however.

Maybe software raid gives just as much reliability these days and it's
mainly a perforance difference?  Perhaps this is a bit too bleeding edge and I should fall back on software raid? Reliability and hot-swap ability are key for this client, not speed.  

Thanks  for the driver Jeff, will try it tonight.

Cheers,
Seth.

[Charlie Brady]

Seth wrote:
 
> > I'm interested to know why you you've gone to so much
> > trouble to build this module?

> Uh, well in my case at least I'm after RAID 1 reliability,
> not raw speed.
...
> Maybe software raid gives just as much reliability these days
> and it's mainly a perforance difference?

It's software raid whichever way you do it. I've seen no evidence that the HPT implementation is more reliable that the standard kernel software RAID, but my gut feeling is to trust the standard kernel more.

> Reliability and hot-swap ability are key for this client, not
> speed.

You don't get host-swap capability with standard kernel software RAID. Do you with the hpt driver? Do you really need it?

Charlie

[Joshua B]

Just to chime in:

I'm going the hardware RAID route because of the CPU and I/O overhead.  Not that there's very much, but the computer makes 1 call to the card and it splits the read/write requests between the RAID 1 disks.  The RocketRaid 133 also has 2 IDE channels.

Do I need all that?  Not right now.  Will I in the future?  Possibly, and the hardware is relatively cheep right now.  Why not include it in the initial install with an eye to the future?

Joshua B

[David Nowakowski]

I have to say that, given the problems I have had getting the HPT404 to work (still unsuccesfull I might ad), I have been tempted to try the softwate implementation of RAID in SME. The problem with that is that I do not want to mirror my OS drive. I have a set of 60 GB data drives that I want to mirror. I do not want to have the OS on my data drive. There is just to much that can go wrong that might choke up the OS. When that happens, I dont want it happening to my data. I have longed believed that seperating the two is a best practice. As it is, if I do something that messes up my install, I just reinstall to my OS drive with no concerns about my data.

just my .02

David Nowakowski

[Jeff]

Charlie... you can find the source at: http://www.acenet-tech.org/cdj/dev-tools/sme56/

[Martijn T]

Jeff, I noticed you managed to compile an Athlon 2.5.18-5 kernel module for the HPT374, however, you also had an Intel 2.5.18-10 version, which i couldn't insert because I have a clean v5.6 (the reference errors)
Do you have an updated version for the older kernel, or could you suggest some instructions howto manage my problem.
I also tried to compile the source from highpoint on a virtual machine, without succes, ofcourse..
(I have a single PII with the Rocketraid404, and I'm not an advanced user)