On Thu, Nov 26, 2009 at 2:28 PM, Florian Haas <florian.h...@linbit.com> wrote: > Andrew and everyone, > > apologies upfront if this is turning into a rant. This has been somewhat > bothering me for a while. > > A bit of backdrop. > > - The docs > (http://www.clusterlabs.org/doc/en-US/Pacemaker/1.0/html/Pacemaker_Explained/apes03s02.html) > have claimed for a while that Pacemaker 1.0.x is compatible with > Heartbeat 2.1.3 (aka Pacemaker 0.6). Thus it ought to be safe to expect > to be able to do rolling upgrades from 2.1.3/2.1.4 to 1.0.x.
Yes, I anticipated it too when I was releasing 1.0.0 In fairness the wiki has been correct since April when I encountered the issue. [snip] > Because, quoting from the documentation, rolling upgrades are "currently > broken between Pacemaker 0.6.x and 1.0.x. If there is sufficient demand, > the work to repair 0.6 -> 1.0 compatibility will be carried out." > > I firmly believe there is sufficient demand. I therefore ask that this > breakage be fixed. Perhaps other Debian users can second that request of > mine. What I don't understand, if such demand exists, is why I'm not hearing more about it. Since 1.0.0 came out over a year ago, I've had exactly 4 people complain about the problem (and only half of those had actually performed an upgrade and encountered the problem). I even explicitly pointed out the problem and asked for people's feedback as to whether it was important. To date that thread has zero replies in 7 months. The occams-razor explanation would seem to be that cluster admins simply don't do rolling upgrades between major versions. Perhaps you can convince lmb to fix it, I think he had thoughts of using that capability. But hey, if hoards of people suddenly turn up saying they simply must have rolling upgrades to 1.0 I will of course work on it myself. _______________________________________________ Pacemaker mailing list Pacemaker@oss.clusterlabs.org http://oss.clusterlabs.org/mailman/listinfo/pacemaker
