1 |
On Sun, 2010-09-12 at 00:52 +0000, Duncan wrote: |
2 |
> Lindsay Haisley posted on Sat, 11 Sep 2010 18:36:21 -0500 as excerpted: |
3 |
> > Ugh!! This sounds like the roadblock I ran into! The problem drive, |
4 |
> > /dev/hda, holds the root filesystem and several others, and didn't show |
5 |
> > up at all in /dev. Since the system also has several SATA drives, I'm |
6 |
> > using the sata_promise kernel module for these, but the SATA system on |
7 |
> > the MB is managed by a Promise chipset, which supposedly implements |
8 |
> > hardware RAID but it's proprietary so I'm just using plain old discrete |
9 |
> > non-RAID mode. |
10 |
> |
11 |
> JBOD (just a bunch of disks) mode. I use it here, too. |
12 |
|
13 |
JBOD - That's like POTS in telco terminology :-) I was trying to think |
14 |
of JBOD but couldn't remember it. |
15 |
|
16 |
> From what you wrote earlier, I thought you had all SATA hard drives but |
17 |
> were depending on udev's compatibility rules to setup hdX symlinks to the |
18 |
> sdX devices, using the hdX symlinks in your fstab. If you're still using |
19 |
> pata and the devices themselves were hdX but are now sdX, that's a |
20 |
> different issue and there's a bit more excuse... |
21 |
|
22 |
No, the MB has an Intel chipset with a standard PATA header/port, but it |
23 |
also has an onboard Promise controller and 4 SATA connectors intended to |
24 |
support RAID. I never could get the Linux hacks with the proprietary |
25 |
Promise RAID controller to work properly, so I just set them up JBOD and |
26 |
incorporated them into Linux RAID-1 arrays. |
27 |
|
28 |
This isn't as bad as our firewall/file server box! I have that set up |
29 |
with EVMS, and it boots from an initrd right into EVMS, so the boot |
30 |
drive is a Linux RAID array. Now EVMS has been abandoned (pity, it was |
31 |
a great idea!) and the box is living on borrowed time as far as upgrades |
32 |
are concerned. It was a daring trick, but it'll probably bite us in the |
33 |
butt eventually! |
34 |
|
35 |
> tho as mentioned really |
36 |
> only if you've been living in a cave or under a rock for some years. |
37 |
|
38 |
I like caves, and under rocks, too ;-) I have a life outside of |
39 |
geek-land and these days it's looking pretty cool. I do fall behind on |
40 |
this stuff, though, as a result. |
41 |
|
42 |
> If you configure your own kernel, you should have run into that when doing |
43 |
> the make oldconfig, and could have adjusted accordingly then. But if you |
44 |
> depend on genkernel... well, let's just say I like to know what changes |
45 |
> are going on with my kernel, and that's one of the reasons I don't use |
46 |
> genkernel. (Tho for all I know, there was a warning when genkernel did |
47 |
> the change too, but I wouldn't know, as I don't use it.) |
48 |
|
49 |
I don't use genkernel either, for the same reasons. I've been |
50 |
configuring my own kernels since Linux kernel 1.something. That kinda |
51 |
dates me, I guess. The only thing genkernel is good for is if you |
52 |
_have_ to build an initrd, but I had to do that by hand anyway when I |
53 |
set up our in-house file server because I had to build EVMS support into |
54 |
it. |
55 |
|
56 |
> > I have rather a conflict here, since I already have a /dev/sda. |
57 |
> |
58 |
> See vvv (below, those are arrows). |
59 |
> |
60 |
> > Is there a HOWTO for using libata-supported kernel components, and |
61 |
> > configuring them in the kernel? |
62 |
> |
63 |
> Someone familiar with the specific hardware you have might know exactly |
64 |
> what order (vvv) the devices would appear in, but I'm not, so it's of no |
65 |
> significance to me and I snipped it. I explain the general situation vvv. |
66 |
|
67 |
Duncan, thanks for your notes and observations. Most of this I already |
68 |
know, but it's always encouraging and helpful to have someone else lay |
69 |
it out. I'll probably take another run at the upgrade when my time |
70 |
permits, which it doesn't right now. I just need to make sure that I |
71 |
have a path to back out of any changes, as I did earlier this week, if |
72 |
things get foobar. I have to be able to get to a stable, bootable |
73 |
desktop at the end of the day, whether it's running the old kernel or a |
74 |
new one. |
75 |
|
76 |
> Among other things, udev should create symlinks for each device UUID/GUID |
77 |
> to the associated name, and if you write down which GUID applies to which |
78 |
> device on your current system, you can use that to figure out which sdX |
79 |
> they end up on with the new layout. |
80 |
|
81 |
I can also look at them with cfdisk and figure it out from the partition |
82 |
layout - which is probably easier. I would assume the SATA drives will |
83 |
be in the same order. The hard drives are also identified by model and |
84 |
what I assume to be the serial number in /dev/disk/by-id - e.g. |
85 |
"ata-WDC_WD360GD-00FNA0_WD-WMAH91500602" - which is, in a pinch, also |
86 |
printed on the drive cases. |
87 |
|
88 |
> Once you've noted the order and figured out which sdX corresponds to which |
89 |
> device, make your rootfs writable as you did before, and change the |
90 |
> corresponding fstab and intermediate layer (lvm/dmraid/mdraid/etc) configs |
91 |
> so the mapping is to the new device names instead of the old ones. |
92 |
|
93 |
grep -R hd /etc/lvm/* says that the only place that "hd" is mentioned |
94 |
(other than in comments) is in /etc/lvm/cache/.cache, which I would |
95 |
assume I could erase and the LVM system would re-create it. "sd" isn't |
96 |
in there at all. I'll also have to edit /etc/mdadm.conf which makes |
97 |
specific references to /dev/sd* partitions by name. |
98 |
|
99 |
> can then reboot, and it should come up as normal. =:^) (If it doesn't, |
100 |
> there's probably a mapping you forgot to change, somewhere.) |
101 |
|
102 |
... and if it fails, I go and cry myself to sleep and work on it the |
103 |
next day ;-) |
104 |
|
105 |
-- |
106 |
Lindsay Haisley | "Humor will get you through times of no humor |
107 |
FMP Computer Services | better than no humor will get you through |
108 |
512-259-1190 | times of humor." |
109 |
http://www.fmp.com | - Butch Hancock |