Hi! I have a question: DRBD can only handle two peers (not three). When planning a dual-primary configuration, one might imageine having a block device on three cluster nodes where CRM picks two of them to run a dual-primary DRBD. Obviously the third unused one will have obsolete data. I guess if one node fails, CRM will pick the so far unsused node to start DRBD and overwrite the obsolete data with current ones. So no data should be lost.
The scenario where data will be lost would be: One node has a problem to keep the data up to date. Then we have two nodes with non-up-to-date data. If the good node now fails before any of the other nodes are synced with current data, some current data will be lost. Did I get the basics right so far? Would onyone recommend such a setup? Has anybody running such a setup in a realistic productive environment? Regards, Ulrich _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
