I'm finalizing the arrangement for a couple of VM servers holding a few VM guests.
My primary concern is that the storage needs to be HA and redundant. Services/VM can afford down/restart time. After months of reading up, concerns over complexity (being a noob admin) and costs of STONITH switches and such, I came up with the following arrangement. Storage Nodes - RAID 5 on physical drives - Two NICs - Export each block device as two iSCSI targets on two different VLANs VM Host -> mdraid raid 5 over mdraid multipath on iSCSI targets -> write intent bitmap on local drive There will be 3 switches in the network. One switch for user access to the VM Host, and two switches just for the two storage VLAN. Multipath is intended to avoid expensive rebuild process on all storage nodes if a network switch fails. Then I can simply monitor for failures of the VM guests, restart them if the host is alive or on another VM host if the host is dead. Each VM guest run their own full set of services (mysql/httpd) so there won't be any need to move services around. So I think that should eliminate the need for any STONITH or am I mistaken? Or are there problems with such a setup that I'm not seeing? Thanks in advance for any advice :) _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
