[Pulp-list] checksumming downloads

[Alan Milligan]

I have some slightly schizophrenic yum servers which depending upon user
agent will return packages or html views.  Seems because Pulp's
python-requests library downloads, rather than urlgrabber (and other
obvious RPM user agents), html is happily downloaded and then treated as
an RPM package on the filesystem. 

Even worse: it's treated by Pulp as *genuine* RPM content: able to be
copied between repo's, published as metadata et al.

It's only when a yum client actually attempts to deploy the package that
checksums do not match.

I have fixed my end now; but since all of this information is available
on feed sync: would it not be worth checksumming the download and taking
action (probably electing to ignore the package) if for whatever reason
a checksum is inconsistent?

Alan