Hello,
I have 4kn (4KiB logical and physical bytes per sector) Advanced Format SAS drives on an Adaptec 81605ZQ hardware RAID controller. I have a logical drive on the controller, that maintains the 4kn formatting, and having it passed to a Windows Server 2016 VM.
Debian sees them as 4kn, and Windows Server 2016 seems them as 4kn when booted natively, but when passed with standard parameters from PVE GUI, they are being read as 512 logical. This causes all sorts off problems with GPT placement, etc... (If I partition in Debian with gdisk or parted, Windows doesn't even see it as a GPT drive. If I convert to GPT, and partition in Windows, Debian sees a corrupted MBR).
I have explored, and I used the "qm showcmd <vmid>" to a text file, and added "physical_block_size=4096,logical_block_size=4096" to the -device 'virtio-blk-pci ... ' section and then just launch the command line manually from debian terminal, and it works properly.
If I add 'physical_block_size=4096,logical_block_size=4096, to the vmid.conf file, I get a schema error, and it won't save them.
If I add the "args: -device '...' " , or "args: -drive '...' -device '...' " to the vmid.conf, it executes the args before reading the q35 config, and there is no pci.0 created yet.
Is there a away to modify the schema to allow for the physical_block_size=4096,logical_block_size=4096 in the vmid.conf file on the virtio0: area, and have it parsed and passed to the -device 'virtio-blk-pci,...' start command?
Alternatively, Is there a way to ensure the args: runs AFTER the reading of the q35 config, thus ensuring pci.0 already exists?
Thank you!
I have 4kn (4KiB logical and physical bytes per sector) Advanced Format SAS drives on an Adaptec 81605ZQ hardware RAID controller. I have a logical drive on the controller, that maintains the 4kn formatting, and having it passed to a Windows Server 2016 VM.
Debian sees them as 4kn, and Windows Server 2016 seems them as 4kn when booted natively, but when passed with standard parameters from PVE GUI, they are being read as 512 logical. This causes all sorts off problems with GPT placement, etc... (If I partition in Debian with gdisk or parted, Windows doesn't even see it as a GPT drive. If I convert to GPT, and partition in Windows, Debian sees a corrupted MBR).
I have explored, and I used the "qm showcmd <vmid>" to a text file, and added "physical_block_size=4096,logical_block_size=4096" to the -device 'virtio-blk-pci ... ' section and then just launch the command line manually from debian terminal, and it works properly.
If I add 'physical_block_size=4096,logical_block_size=4096, to the vmid.conf file, I get a schema error, and it won't save them.
If I add the "args: -device '...' " , or "args: -drive '...' -device '...' " to the vmid.conf, it executes the args before reading the q35 config, and there is no pci.0 created yet.
Is there a away to modify the schema to allow for the physical_block_size=4096,logical_block_size=4096 in the vmid.conf file on the virtio0: area, and have it parsed and passed to the -device 'virtio-blk-pci,...' start command?
Alternatively, Is there a way to ensure the args: runs AFTER the reading of the q35 config, thus ensuring pci.0 already exists?
Thank you!
