Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Wed, 2008-02-20 at 22:44 +0000, Massimo Mongardini wrote: > Rois, > thanks for this > > rois wrote: > > On Tue, 2008-02-19 at 22:22 +0000, Massimo Mongardini wrote: > > > > > Rois, > > > I don't have such sections on my drbd.conf is that relevant to > > > drbd 8.x perhaps. My versions are drbd 0.7.24 and heartbeat 2.1.2-3 > > > I planned a downtime on Saturday during which I will try and see if I > > > can reproduce the fault again. > > > cheers > > > Massimo > > > > > > > > I'm using drbd 8 and don't know anything about 7 but part of what is in > > these sections is the outdate-peer info. One of my initial problems > > when I was setting up was that a failure would not fail over because the > > secondary was being outdated (and thus not able to become primary) when > > the primary failed. Someone who is more familiar with drbd 7 will have > > to help you if this section is not important to drbd7. > > > I can fail over, and did plenty of tests with programs writing on the > shared nfs and rebooting the primary and they were all satisfactory, > but for Murphy's law (and perhaps cosmic rays!) for once that it had > to do it in production ... it failed! When you say you did fail over tests do you mean you did "heartbeat standby" or did you actually pull the power plug or all the network cables? I've noticed that the drbd/HA acts different when you do nice fail over test (rebooting the machine or doing "heartbeat standby") as apposed to "down right nasty" fail over tests (pull the power plug or all network cables at once to simulate something very bad happening.) I'm fairly paranoid about how stuff works so I like pushing the limit. Rois