Hi there.
My simple structure: two nodes each with esxi 6.7 and local storage. Datacore [sansymphony] is serving virtual disks to esxi through iscsi. Each node has a datacore vm with its software installed. Every time I reboot a node or I turn off both nodes, after reactivating datacore services I need to perform a manual iscsi storage adpater rescan in order get my vm's back. Without this operation, all datastores served by datacore are unaccessible. Datacore support told me there is nothing they can do as it's a esxi issue . This is their answer "
We would expect the Host to be able to detect when a disk or device's path is available automatically.
We have no control over what the Host does (or the Hypervisor were we run inside the VM). Any time the vDisk mirror status changes (either from healthy to failed, failed to recovery or recovery to up) we will issue both SCSI and ALUA commands up via the FE ports for the Host to detect (either on the local path or the 'other' DataCore Server's path) so we would expect these should force the Host to do its own device rescan.
Also note there is usually a difference between an iSCSI Initiator logged into our Target and the iSCSI Initiator/Host detecting a device.
Path down in this case means that the initiator is not even logging into our target, which should happen even if there is no mirror device connected or ready.
When we are in recovery we will block access to one side of the mirror but we do not stop any host from logging into the FE port - they are different things (device discovery and Port negotiation).
So I would say no this is not usual behaviour but it is not something we can or do control. The Host should be detecting our Target driver and then logging in, and then we'd expect it to query for any devices. It seems to me that the port(s) on this Host are not even logging back in in."
Any suggestion?
thanks in advance
Max
Italy
