| 1 |
-----BEGIN PGP SIGNED MESSAGE----- |
| 2 |
Hash: SHA1 |
| 3 |
|
| 4 |
Beso wrote: |
| 5 |
>> can i use raid even if i got a single hd and a non raid board?! i think |
| 6 |
>> i missed this thing. i knew that i could use raid on 2 separate disks of |
| 7 |
>> the same ammount and only if i had a raid compatible board (with |
| 8 |
>> hardware or software) but i didn't know that you could use it also on a |
| 9 |
>> single disk. |
| 10 |
> |
| 11 |
|
| 12 |
The simple answer is no - you can't use raid with one hard drive. The |
| 13 |
whole point of raid is to provide increased transfer rate and/or |
| 14 |
redundancy (emphasis on the latter) by combining multiple drives together. |
| 15 |
|
| 16 |
You do not need any special hardware to make raid work on linux (this is |
| 17 |
software raid). In fact, unless it is a $1000 adaptec raid controller |
| 18 |
with battery backup that you're using I'd AVOID using any special |
| 19 |
hardware. My motherboard has "built-in raid support" and I don't use it |
| 20 |
for my two raid-5 arrays or any of my raid-1s. The cheap hardware |
| 21 |
support provides little benefit and can cause problems (and these |
| 22 |
solutions are almost always inflexible). |
| 23 |
|
| 24 |
If you have only one hard drive just set up a boot partition (small), a |
| 25 |
root partition (bigger - mine is 1GB, but you could go a little smaller |
| 26 |
or larger), and then have one big partition assigned to lvm and then |
| 27 |
split that up to handle everything else. |
| 28 |
|
| 29 |
If at a later date you decide to install more hard drives and go with |
| 30 |
raid the lvm partitions will be trivial to migrate. Your boot will also |
| 31 |
be easy - it isn't in use while the system is up. The only pain will be |
| 32 |
your root partition, and that will be mitigated by the fact that there |
| 33 |
will be next to nothing on it. |
| 34 |
|
| 35 |
> |
| 36 |
>> i tried to copy the system some time ago and found out that there are |
| 37 |
>> files in /dev and /tmp or /var/tmp that have an enormous dimension. i |
| 38 |
>> have left them behind and then got an unusable system for some reason. |
| 39 |
>> the copy i had was from a livecd with the cp -p to preserve ownership |
| 40 |
>> and permission. |
| 41 |
>> for what i know from /dev i have only to get /dev/null and /dev/console |
| 42 |
>> and let all others devices be created by udev. from /tmp instead i |
| 43 |
>> should not copy anything and from /var/tmp i should copy only the |
| 44 |
>> ccache. are my suppositions correct? |
| 45 |
> |
| 46 |
|
| 47 |
If you copy files with cp - use the -a flag to make it not dereference |
| 48 |
links/devices/etc. "cp -a /dev/zero /tmp/zero" works just fine. |
| 49 |
|
| 50 |
I would recommend making /tmp and /var/tmp tmpfs filesystems. It |
| 51 |
greatly improves performance and you shouldn't be storing anything in |
| 52 |
these directories for longer than a reboot. I also make myself another |
| 53 |
tmp directory on a regular hard drive for "junk" and have tmpreaper keep |
| 54 |
it clean - but it gets rare use. In any case, even if you don't use |
| 55 |
tmpfs I wouldn't bother copying them - losing your ccache isn't that big |
| 56 |
a deal. |
| 57 |
|
| 58 |
As somebody else pointed out, udev can take care of most of /dev, but |
| 59 |
you do need at least a few devices there for bootup. I don't know which |
| 60 |
ones offhand, but you could just extract a stage 1 tarball someplace and |
| 61 |
copy it's device directory for a core set of files. |
| 62 |
-----BEGIN PGP SIGNATURE----- |
| 63 |
Version: GnuPG v1.4.7 (GNU/Linux) |
| 64 |
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org |
| 65 |
|
| 66 |
iD8DBQFHEOOjG4/rWKZmVWkRAlIvAJ0T4ZNKy2BovkbyzSLAigdYq3/EQQCeNygd |
| 67 |
1HGLnNxkh6RkmWaKa7KT07g= |
| 68 |
=Dd3P |
| 69 |
-----END PGP SIGNATURE----- |
Archives