1 |
Drew Tomlinson wrote: |
2 |
|
3 |
> On 12/10/2005 1:17 PM Stroller wrote: |
4 |
> |
5 |
>> |
6 |
>> On Dec 10, 2005, at 5:08 pm, drew@××××××××××××××.net wrote: |
7 |
>> |
8 |
>>> |
9 |
>>> I have a system on an Abit motherboard with the Nvidia GeForce 4 |
10 |
>>> chipset. |
11 |
>>> There are two SATA disks in a hardward stripe configuration using the |
12 |
>>> controller built in to the motherboard. |
13 |
>>> ... |
14 |
>>> I booted the latest Knoppix dated 9/23/05 and see two |
15 |
>>> icons on the desktop for my drives. One for sda the other for sdb. |
16 |
>>> I can |
17 |
>>> not mount either. I assume this is because Knoppix is seeing each |
18 |
>>> drive |
19 |
>>> individually instead of the one logical striped drive it is. |
20 |
>> |
21 |
>> |
22 |
>> |
23 |
>> I'm no expert on this, but I believe that many ATA "hardware" RAID |
24 |
>> arrangements in fact just use their Windoze drivers to do software |
25 |
>> RAID. I'd do some research via Googling the chipset &/or board's |
26 |
>> model number if I were you. |
27 |
> |
28 |
> |
29 |
> |
30 |
> I haven't found anything yet but then I haven't looked real hard. |
31 |
> However I suspect this does not rely on any Windows drivers as the |
32 |
> controller is managed long before Windows boots. Just after POST and |
33 |
> before the OS starts, a brief message showing the controller is |
34 |
> displayed. By pressing F10, I can manage my stripe. Much like I see |
35 |
> most SCSI cards. |
36 |
|
37 |
|
38 |
Stroller is right on this one. Most built-in motherboard 'RAID chips' |
39 |
and inexpensive RAID cards are simply software RAID with a BIOS front |
40 |
end that can be used to create/manage RAID volumes composed of attached |
41 |
disks. The firmware for these RAID chips/cards simply scans each disk |
42 |
for its proprietery RAID metadata, and presents any found volumes as a |
43 |
drive to the BIOS. They also often detect problems with the arrays |
44 |
(missing components, etc) and present the user with boot time options to |
45 |
deal with these situations (boot in degraded mode, replace missing |
46 |
component and sync, etc). But the actual block-by-block RAID operations |
47 |
(writing mirror blocks, computing/reading/writing parity data, degraded |
48 |
operation, etc) are done in a device driver under the OS, once booted. |
49 |
The CPU is doing all the RAID heavy lifting with these arrangements. I |
50 |
see these sorts of RAID solutions as merely glorified IDE or SATA |
51 |
controllers with some advanced firmware in front of them. |
52 |
|
53 |
True hardware RAID do these block-by-block operations on a dedicated |
54 |
controller, often with some sort of NVRAM write-behind/read-ahead cache |
55 |
between the OS and the volume, which can survive sudden power |
56 |
loss/crashes, etc. The RAID volumes presented to the boot time |
57 |
environment and OS look like regular drives. The RAID controller |
58 |
hardware does the RAID heavy-lifting (parity computation, re-syncs, |
59 |
etc), offloading them from the CPU. Obviously these hardware RAID |
60 |
solutions are more sophisticated than the cheap software RAID |
61 |
arrangements, which is why presently, you won't find one for less than |
62 |
about $400 (USD), where you can find the software RAID cards for < $100 |
63 |
(USD). |
64 |
|
65 |
Both types seem to present a similar user interface to the user, which |
66 |
is cause for confusion. Also, the vendors of the cheap RAID solutions |
67 |
don't go out of their way to inform the customers of the differences |
68 |
between their stuff and the hardware RAID solutions, of course, which is |
69 |
more cause for confusion. :-) |
70 |
|
71 |
|
72 |
>>> Is there some magic I can perform at the boot prompt to get Knoppix |
73 |
>>> to see the |
74 |
>>> two individual drives as one logical striped drive? I can't recover |
75 |
>>> the data |
76 |
>>> from booting Windows because it's all screwed up and reboots itself |
77 |
>>> shortly |
78 |
>>> after logon. |
79 |
>> |
80 |
>> |
81 |
>> |
82 |
>> If my guess is correct then the best thing might be to install |
83 |
>> Windows on a spare drive & boot from that to see the RAID as one. You |
84 |
>> might try booting with a Windows CD & see if the RAID is recognised |
85 |
>> as a single partition... if you get the option to do a repair install |
86 |
>> you _should_ be able to get an at-least-mostly-working Windows |
87 |
>> install & all your data intact. Recover the data to a portable drive |
88 |
>> & format. |
89 |
> |
90 |
> |
91 |
> |
92 |
> Thanks. I tried an overlay install again and things seem to be going |
93 |
> well. Copying data now. |
94 |
> |
95 |
> Thanks for your ideas! |
96 |
> |
97 |
> Drew |
98 |
> |
99 |
|
100 |
As mentioned by another poster in this thread, there is also the |
101 |
somewhat too generically named 'dmraid'. (When I first stumbled across |
102 |
this, I thought it was some LVM2 native replacement for the good 'ole |
103 |
Linux RAID [md] devices. I personally think it should be given a more |
104 |
specific name like 'metaraid' or 'omniraid' or soemthing like that, but |
105 |
I digress). |
106 |
|
107 |
You may have been able to use this to mount your striped volume. |
108 |
|
109 |
From my admittedly brief reading, dmraid appears to be a metadata |
110 |
agnostic device mapper based software RAID driver. That is, it can |
111 |
understand and operate RAID volumes created using many vendors' software |
112 |
RAID chips and cards and their proprietary metadata formats by itself, |
113 |
without the need to install drivers from individual vendors. Pretty cool! |
114 |
|
115 |
However, the userspace tools for dmraid are presently keyword masked in |
116 |
gentoo, and the somewhat sparse documentation for dmaid seems to |
117 |
indicate that it's not quite ready for prime time yet. Looks |
118 |
interesting though. |
119 |
|
120 |
|
121 |
-- |
122 |
+---------------------------------------------------------------------------+ |
123 |
| Jim Burwell - Sr. Systems/Network/Security Engineer, JSBC | |
124 |
+---------------------------------------------------------------------------+ |
125 |
| "I never let my schooling get in the way of my education." - Mark Twain | |
126 |
| "UNIX was never designed to keep people from doing stupid things, because | |
127 |
| that policy would also keep them from doing clever things." - Doug Gwyn | |
128 |
| "Cool is only three letters away from Fool" - Mike Muir, Suicyco | |
129 |
| "..Government in its best state is but a necessary evil; in its worst | |
130 |
| state an intolerable one.." - Thomas Paine, "Common Sense" (1776) | |
131 |
+---------------------------------------------------------------------------+ |
132 |
| Email: jimb@××××.cc ICQ UIN: 1695089 | |
133 |
+---------------------------------------------------------------------------+ |
134 |
| Reply problems ? Turn off the "sign" function in email prog. Blame MS. | |
135 |
+---------------------------------------------------------------------------+ |
136 |
|
137 |
-- |
138 |
gentoo-user@g.o mailing list |