-
Notifications
You must be signed in to change notification settings - Fork 85
show summary of image sizes at end of tests #162
Comments
At the end of the tests, show a summary of the sizes of all the images and initrds along with their sizes. Fixes kata-containers#162. Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
At the end of the tests, show a summary of the sizes of all the images and initrds along with their sizes. Fixes kata-containers#162. Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
Currently images size is set to always be 512MB: Line 13 in 93ad049
osbuilder/image-builder/image_builder.sh Line 186 in 93ad049
|
At the end of the tests, show a summary of the sizes of all the images and initrds along with their sizes. Fixes kata-containers#162. Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
At the end of the tests, show a summary of the sizes of all the images and initrds along with their sizes. Fixes kata-containers#162. Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
At the end of the tests, show a summary of the sizes of all the images and initrds along with their sizes. Fixes kata-containers#162. Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
At the end of the tests, show a summary of the sizes of all the images and initrds along with their sizes. Fixes kata-containers#162. Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
iirc, we do still round up to a nice boundary though on the actual .img file size. What we really want to see is more like a 'df' of what is inside the image (so we could loop mount and have a peek maybe), or really really something like a PSS measure running a single |
@grahamwhaley: how is PSS affected by the rootfs size? Is it because we cache in memory the guest filesystem? |
At the end of the tests, show a summary of the size of all the rootfs's, images and initrd's. Fixes kata-containers#162. Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
@marcov tbh, I'm not sure it is. It might be that QEMU has to cache more or construct more (larger) structures to do the mappings etc. It is probably more likely that the costs on the host side will be in the host side kernel (page tables etc.) maybe, which we'd see with the system level density test. But, that test is probably not going to give a stable result when running a single container. |
At the end of the tests, show a summary of the size of all the rootfs's, images and initrd's. Fixes kata-containers#162. Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
At the end of the tests, show a summary of the size of all the rootfs's, images and initrd's. Fixes kata-containers#162. Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
It would be useful to see a formatted list of the images and initrds at the end of the test, showing the sizes of each.
The text was updated successfully, but these errors were encountered: