1 |
thegeezer <thegeezer@×××××××××.net> writes: |
2 |
|
3 |
>> Guess what, I still haven't found out how to actually back up and |
4 |
>> restore a VM residing in an LVM volume. I find it annoying that LVM |
5 |
>> doesn't have any way of actually copying a LV. It could be so easy if |
6 |
>> you could just do something like 'lvcopy lv_source |
7 |
>> other_host:/backups/lv_source_backup' and 'lvrestore |
8 |
>> other_host:/backups/lv_source_backup vg_target/lv_source' --- or store |
9 |
>> the copy of the LV in a local file somewhere. |
10 |
> |
11 |
> agreed. you have two choices, you can either use dd and clone the LV |
12 |
> like a normal partition. |
13 |
> alternatively you can use split mirrors and i do this to clone up |
14 |
> physical devices: |
15 |
> |
16 |
> 1. make a mirror of the lv you want to copy to /dev/usb1 |
17 |
> 2. # lvconvert --splitmirrors 2 --name copy vg/lv /dev/usb1 |
18 |
> |
19 |
> in 2 it says |
20 |
> " split the mirror into two parts |
21 |
> give the new version the name 'copy' |
22 |
> leave this on the pv /dev/usb1 " |
23 |
> |
24 |
> you then need to remove it if you want from your voume group |
25 |
|
26 |
And then you have a copy? |
27 |
|
28 |
>> Just why can't you? ZFS apparently can do such things --- yet what's |
29 |
>> the difference in performance of ZFS compared to hardware raid? |
30 |
>> Software raid with MD makes for quite a slowdown. |
31 |
> sorry but that's just not true, if you choose the correct raid level and |
32 |
> stripe it can easily compete, |
33 |
|
34 |
No, it can not. You can use the same raid level and stripe size with |
35 |
the same disks, once with software raid (md), once with hardware raid. |
36 |
When there's disk activity going on, you will notice an annoying |
37 |
sluggishness with the software raid which hardware raid just doesn't |
38 |
have. |
39 |
|
40 |
> and be more portable (you don't have to find the identical raid card |
41 |
> if the raid card goes bang); |
42 |
|
43 |
That depends on what cards you use. HP smart arrays are supposed to be |
44 |
compatible throughout different models. |
45 |
|
46 |
> many raid card i would argue are even underpowered for their required |
47 |
> iops |
48 |
|
49 |
When you require more iops than your hardware delivers, you need to |
50 |
upgrade the hardware. Do you have some number to show that md raid |
51 |
performs better, without the sluggishness, than a decent hardware raid |
52 |
controller with otherwise identical hardware? |
53 |
|
54 |
|
55 |
-- |
56 |
Again we must be afraid of speaking of daemons for fear that daemons |
57 |
might swallow us. Finally, this fear has become reasonable. |