After discussion, there are a few things to investigate. Tops would be trying to reproduce this with si 3.0 rather than the snappy branch of 2.5. I am putting together a PPA version of 3.0 for James and Michael to test.
If it's reproducible there, then some further thoughts:
* Are we getting accurate HEAD information from system-image.ubuntu.com?
* Do the file systems in the image json match the HEAD calculated total size?
* Is it a problem with the math for calculating the % downloaded so far?
Obviously it would be helpful to come up with a reproducible test case.
After discussion, there are a few things to investigate. Tops would be trying to reproduce this with si 3.0 rather than the snappy branch of 2.5. I am putting together a PPA version of 3.0 for James and Michael to test.
If it's reproducible there, then some further thoughts:
* Are we getting accurate HEAD information from system- image.ubuntu. com?
* Do the file systems in the image json match the HEAD calculated total size?
* Is it a problem with the math for calculating the % downloaded so far?
Obviously it would be helpful to come up with a reproducible test case.