Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Christof Amelunxen wrote: >Hi Clive, > >thanks for your reply. > >Am Mo, den 31.10.2005 schrieb Clive Eisen um 16:17: > > >>>we plan to use DRBD and heartbeat to cluster an Informix DB and we would >>>like to use raw devices to store the data on. >>> >>>Has anybody ever tried using Linux raw devices on top of DRBD or could >>>one of the technical experts around tell me if this is possible at all? >>> >>> >>> >>> >>> >>I have done it as a demo, not in production. >> >>It works fine with raw and heartbeat although depending on your distro >>you may have to hack the 'raw' script to actually work >> >>http://groups.google.co.uk/group/novell.support.suse.linux.professional/browse_thread/thread/62b8da97916987dd/b11209c7c384fa9e?lnk=st&q=%22clive+eisen%22+raw&rnum=4&hl=en#b11209c7c384fa9e >> >> > >Is it necessary to have the raw devices be managed by heartbeat? I was >thinking about binding the raw devices statically to the drbd >major/minor numbers. > > > IIRC I couldn't bind raw on the drbd secondary, only on the primary, which IMHO makes sense. Therfore raw had to be managed by heartbeat >>Later informix versions open the chunks as 'raw' without you having to >>go through these hoops, and will also use KAIO if possible >> >> > >But does that make any difference in practice? At least for me, the >"drbd administrator"? > > > Yes because you don't have to have raw in your haresourses - it's no big deal, but just one less thing to get working -- Clive