1 |
Brian, |
2 |
first of all thanks for your answer |
3 |
|
4 |
Il giorno lun, 18/05/2009 alle 09.29 -0500, Brian Kroth ha scritto: |
5 |
> I remember looking at this a while ago as well and being fairly |
6 |
> confused. I'm fairly certain that lvm2 with the use flags you mention |
7 |
> is the way to go, so you might have to unmask them in your |
8 |
> /etc/portage/profile/use.mask (add -clvm, -cman). |
9 |
> |
10 |
> OCFS2 generally uses it's own different cluster stack than CMAN (which |
11 |
> uses openais). Since I believe 2.6.26 you can use the other one, but I |
12 |
> haven't found it to be as stable. |
13 |
|
14 |
I didn't understand here what you mean. Which one is "the other one"? |
15 |
(maybe it's just my english :P ) |
16 |
|
17 |
> It also requires quite a bit of extra |
18 |
> configuration for proper fencing, whereas that's built into the OCFS2 |
19 |
> stack. Since you're doing this with DRBD it sounds like you're trying |
20 |
> to do it on the cheap so proper fencing probably isn't available |
21 |
> anyways. |
22 |
|
23 |
Yes, you're right. The only way I have to get a sort of STONITH (if it |
24 |
can be used, I've still to check this) is IPMI which is working well so |
25 |
far. |
26 |
|
27 |
> If you aren't doing anything with LVM that needs to be cluster aware |
28 |
> (eg: just a concatenation of multiple PVs, no striping) then it should |
29 |
> be just fine without CLVM. The key is that you /must/ take all but one |
30 |
> node offline (umount; vgchange -an) before you do any LVM operations |
31 |
> such as adding more extents, else they'll have an inconsistent view of |
32 |
> the VGs/LVs. This also isn't technically supported but I've run it |
33 |
> without problems for a year or so now. |
34 |
|
35 |
The problem is that a node will use (mount) some LVs and the other one |
36 |
some others. So I can't do the fast way as you suggest, as I need to |
37 |
stop all services that use data on LVs before unmounting. |
38 |
|
39 |
I'll try to unmask those flags and re-emerge lvm2. |
40 |
|
41 |
Thanks, |
42 |
Giacomo |