Our Highly Available iSCSI Solution using SCST, Pacemaker and Corosync

[@eros6182]

FINALLY! I have always been missing the iscsi part for all other "webui" solutions except truenas, and now you´re making it a part of houston AND with clustering so i will definitively migrate my stuff over to houston. THANK YOU FOR DOING THIS!

[@kalam564]

Great video. It would be cool to see how a Windows Server VM on ESXi with its drives on a VMDK handles an iSCSI failover, especially in that longer worst case wait period. Looking forward to seeing this in Houston.

[@johnharrison712]

Is it feasible to deploy two Dell servers, each installed with the requisite 45 software, interconnected via a 10GB cable to establish a high-availability (HA) setup? In this configuration, if Host A experiences a failure, Host B would seamlessly take over, thereby ensuring iSCSI HA functionality.

[@MarkRouleau]

I would like to see this with VMs running on the iSCSI device. Do they crash? Corrupted?

[@mitcHELLOworld]

haha well that wouldn't be a very good HA iSCSI solution, now would it? haha all jokes aside, this most definitely does not cause VMs to crash and definitely no corruption. as we are using synchronous IO only data safely committed to disk will send out an ACK to the VM. Also, because its backed by Ceph - Ceph itself fully uses synchronous IO which means the write must be committed to all nodes within the PG before the ACK. This has been tested heavily internally before rolling it out to customers, and is very similar to other enterprise storage solutions.