Hi,
I have a three node PVE cluster where all three PVE nodes are also CEPH nodes. Each node has two NVMEs (that altogether form one pool) and one HDD (that altogether form another pool). My VMs live on the NVME pool. When a VM needs an additional disk for data, that data disks lives on the HDD pool.
When I destroy a VM, its disks get removed from their storage. Normally, when a virtual disk - which generally is only a file on the host - is removed, this does not take long. It is my understanding that the disk/file is only marked as deleted and may therefore later be overwritten but is not actually overwritten at this time.
Why does removing the virtual disks sitting on CEPH take so long? PVE has been at it for 20 minutes now and only managed to remove one 15GB disk on the NVME pool and 20% of a (50% full) 1TB disk on the HDD pool. Does it actually overwrite each byte with zero now or what is going on? Can this process be expedited somehow?
Thanks!
I have a three node PVE cluster where all three PVE nodes are also CEPH nodes. Each node has two NVMEs (that altogether form one pool) and one HDD (that altogether form another pool). My VMs live on the NVME pool. When a VM needs an additional disk for data, that data disks lives on the HDD pool.
When I destroy a VM, its disks get removed from their storage. Normally, when a virtual disk - which generally is only a file on the host - is removed, this does not take long. It is my understanding that the disk/file is only marked as deleted and may therefore later be overwritten but is not actually overwritten at this time.
Why does removing the virtual disks sitting on CEPH take so long? PVE has been at it for 20 minutes now and only managed to remove one 15GB disk on the NVME pool and 20% of a (50% full) 1TB disk on the HDD pool. Does it actually overwrite each byte with zero now or what is going on? Can this process be expedited somehow?
Thanks!