Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
/ 2004-07-11 15:04:07 -0700 \ Steve M Bibayoff: > Hello, > > Lars Ellenberg <Lars.Ellenberg at linbit.com> wrote: > / 2004-07-02 20:15:55 +0200 > \ Stefan Andersson: > > > Would it be possible to use GFS on top of DRBD to get write > > support at > > > both nodes? I.e. does drbd replicate writes at silas to paul? > > > > not yet. but GFS support is on our list. > > somewhen... > > depending on time, human resources, and cash-flow :) > > Is anybody intimately familiar w/ both DRBD and GFS and knows > what would be involved to get both working together to get > write on both nodes working? I know this has been discussed > before, but this feature seems to get being pushed back. > Wasn't it stated at one time for inclusion into the 0.7 > release? well, surely not into 0.7.0. I guess we stated "somewhen during the 0.7 cycle" But actually, concurrent access is likely a reason to bump the release number again... > I know this is a huge problem to solve. And I believe > Lars (sorry if I mis-attribute the email) wrote an email > awhile back describing the type of file locking and granduarity > needed to achieve this goal. Which went way over my head > and I couldn't think of any simple process to achieve > a solution to most of the problems. The main reason is that we need to prepare for "disconnect/reconnect" somehow, and define which part of the device needs to be synchronized in what direction... or ... well, no details now. but there are several ways to do it, and we need to agree on one of them. and then implement, and test. ETA spring/summer 2005, maybe sooner, maybe later :) If someone (me?) works full time on this certainly sooner ... But I think there are some other architectural things that need to be sorted out first, otherwise code may become unmaintainable. Lars Ellenberg