add deepseek and gemma results

1 files changed, 48 insertions, 0 deletions

diff --git a/results/classifier/gemma3:12b/files/688 b/results/classifier/gemma3:12b/files/688
new file mode 100644
index 00000000..6e74f650
--- /dev/null
+++ b/results/classifier/gemma3:12b/files/688
@@ -0,0 +1,48 @@
+
+Shrinking an image with qemu-img  does not reduce image file size
+Description of problem:
+I have a macOS 10.9 VM using a qcow2 image that was 151GB. The image was originally converted from a VMware image with:
+```
+qemu-img convert macOS-10.9.vmdk -O qcow2 -o preallocation=falloc macOS-10.9.qcow2
+```
+This resulted in `macOS-10.9.qcow2` being 151GB big:
+```
+$ du -h macOS-10.9.qcow2 
+151G     macOS-10.9.qcow2
+```
+After reducing the filesystem size from within macOS to 25GB with DiskUtil, I shut down the VM and resized the image to 30GB with:
+```
+qemu-img resize -f qcow2 --shrink macOS-10.9.qcow2 30G
+```
+This succeeded. However, the file still consumes 151GB of space:
+```
+$ du -h macOS-10.9.qcow2 
+151G     macOS-10.9.qcow2
+```
+Even though `qemu-img info` shows:
+```
+$ qemu-img info macOS-10.9.qcow2 
+image: macOS-10.9.qcow2
+file format: qcow2
+virtual size: 30 GiB (32212254720 bytes)
+disk size: 30 GiB
+cluster_size: 65536
+Format specific information:
+    compat: 1.1
+    compression type: zlib
+    lazy refcounts: false
+    refcount bits: 16
+    corrupt: false
+    extended l2: false
+```
+The size inside the VM is also reported as being 30GB.
+
+The whole point of resizing that image was to free up disk space on the host. But this doesn't seem to be happening.
+
+My filesystem is ext4.
+Steps to reproduce:
+1. Create a vmdk image with `qemu-img create -f vmdk test.vmdk 5G`
+2. Convert the vmdk image to qcow2 with `qemu-img convert test.vmdk -O qcow2 -o preallocation=falloc test.qcow2`
+3. Shrink the new image with `qemu-img resize -f qcow2 --shrink test.qcow2 3G`
+
+The resulting `test.qcow2` file should be 3GB, but it's not. It's 5GB.