Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
/ 2006-08-18 15:56:30 +0200 \ Christiaan den Besten: > Hi ! > > Like many other (well, at least since the last two weeks since > I've been following the list .. <grin>) I am also very > interested in the 0.8 brance due to its primary/primary role > ability. > > I have succesfully tested with GFS and OCFS2. Both seem te be > working just fine ... Doing a bonnie++ on both nodes at the same > time (but in different subdirectory's on the drbd0) seems to > work without any problem. > > But ..... rebooting a node while the other stays active ... or > faking it's crash tends to lead upto a split brain as others > have noticed. Even if the 'broken one' has done nothing on its > disks. Most of the times only a full re-sync seemed to work. I > have not tried the "drbdadm -- --discard-my-data connect all" > though, will do that at once ! > > Would it be very hard to indeed 'detect' a broken (or rebooted) > node and let it catch-up with the running available primary? the algorithm we had in mind is able to do that. the implementation is currently not, aparently. we are going to work that out. as soon as we are confident that the remaining issues with single-primary are resolved. -- : Lars Ellenberg Tel +43-1-8178292-0 : : LINBIT Information Technologies GmbH Fax +43-1-8178292-82 : : Schoenbrunner Str. 244, A-1120 Vienna/Europe http://www.linbit.com : __ please use the "List-Reply" function of your email client.