[ovirt-users] Re: VM hangs when expanding direct LUN

Hi Jorge, In short our environment is set up like this: oVirt Cluster (stand alone cluster manager) contains 5 physical servers (nodes), each node contains up to 5 VM (Linux/Windows) from a storage perspective this is set up as a "host-cluster" as IBM calls it, meaning that all 5 nodes have access to all volumes in the cluster, for instance if node 1 hosts VM 1 and 2, node 2 hosts VM 3 and 4 and so on, every node has access to all volumes, and the volumes can be listed on each node using multipath -ll, regardless of the volume being used by a VM on the specific node or not. When expanding a volume or a set of volumes, this is done on the storage level (block level), when this is done the volume is located on the specific VM and a refresh LUN is issued from the GUI. In some cases this works fine in other cases it crashes the VM... Eventlog from manager: Jan 29, 2025, 8:28:31 AM - VM xxxxx is down with error. Exit message: Lost connection with qemu process. Jan 29, 2025, 8:27:45 AM - VM xxxxx is not responding. Jan 29, 2025, 8:27:16 AM - Status of host yyyyyy was set to Up. Jan 29, 2025, 8:27:15 AM - VM xxxxx is not responding. Jan 29, 2025, 8:26:57 AM - Host cluster zzzzzz was updated by system Jan 29, 2025, 8:26:49 AM - Host yyyyyy is not responding. Host cannot be fenced automatically because power management for the host is disabled. Jan 29, 2025, 8:26:49 AM - Direct LUN synchronization failed. Jan 29, 2025, 8:26:49 AM - VDSM yyyyyy command ExtendVmDiskSizeVDS failed: Message timeout which can be caused by communication issues Jan 29, 2025, 8:26:30 AM - VM xxxxxx is not responding. Jan 29, 2025, 8:26:00 AM - VM xxxxxx is not responding. Jan 29, 2025, 8:24:45 AM - VM xxxxxx is not responding. Jan 29, 2025, 8:23:40 AM - Direct LUN synchronization started. From here the only way to get the VM to respond is to log into the node where the VM resides and through virsh do a destroy xxxxxx /John