Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On 9/1/08, Jerry Amundson <jamundso at gmail.com> wrote: > On Mon, Sep 1, 2008 at 8:25 PM, Alexandre Biancalana > <biancalana at gmail.com> wrote: > > Hi list, > > > > I know that this subject was discussed before but I don't find any > > definitive answer to this question, so I hope that this message can > > help people finding this in future. > > > Doesn't matter. Most people seem to find the *official* answer, which > *official* people refuse to change, yet many times just does not work. Hi Jerry ! I just want something that work.... > > > > I installed Debian 4 (etch), Xen 3.2 and DRBD 8.0.13 (from > > backports). The objective is to build a two-node Linux HA to > > virtualize some legacy services and development machines. > > > > My question is about block-drbd script, when I try to boot a virtual > > machine using a drbd disk resource like this: > > > > disk = [ 'drbd:app01,hda,w' ] > > > Why did you use that? Because theoretically when I use drbd: instead phy: Xen call block-drbd script that handles the necessary DRBD resource state transitions, right ? > > When I use the drbd device directly: > > > > disk = [ 'phy:/dev/drbd0,hda,w' ] > > > > everything works great, but I don't want to write wrapper scripts to > > handle live migration. > > > Of course phy: works great. Fortunately, with allow-two-primaries;, > and other controls, DRBD 8.x can do Xen live migration, even using > phys: ... I already read that, but if everything (including state transition and consequently live migration) work directly with phy: why block-drbd (drbd:) exists at all ??? Regards,