On May 26, 2017 3:04:03 AM EDT, Adrian Bunk <b...@debian.org> wrote: >On Fri, May 26, 2017 at 05:46:22AM +0000, Scott Kitterman wrote: >> >> >> On May 26, 2017 12:30:17 AM EDT, Neil Williams <codeh...@debian.org> >wrote: >> >On Fri, 26 May 2017 04:11:49 +0000 >> >Scott Kitterman <deb...@kitterman.com> wrote: >> > >> >> I don't see any way to completely resolve this before stretch >> >> releases other than removing lava-server. >> > >> >This is a *django* bug! There is no evidence that lava-server is >> >responsible for this - it just has to use multiple django apps. Any >> >other system outside Debian using multiple apps will suffer the same >> >bug. >> > >> >Note: once this bug occurs, unless the user is *clearly* warned to >look >> >for and install 1.8, *data loss* will occur. We have already had one >> >user lose data covering 6 months of tests due to this error. The >> >process must halt when the exception is seen, the app will be down >and >> >unusable and users will get a 503. >> >> Sounds like a great input for the release notes. You should submit >it. >> >> If you'd filed this bug months ago, back when you found it, maybe >something could have been done. Given short time before release, >there's not a lot of options. >> >> It really doesn't matter whose fault the bug is. > >Let's get constructive: > >If it is agreed that this is a bug in Django, then you or Neil or the >Django maintainers should forward it to upstream for getting a fix from >upstream. > >If there is no fix for Django available in time for the stretch >release, >the problem should be noted in the release notes (especially if it >might >also affect other users of Django) *and* the fix should be delivered to >users in a stretch point release.
It's not an upgrade path that upstream supports. They aren't going to 'fix' anything for this even if they agree it's a Django problem (I'm not saying it is or isn't, it doesn't matter at this point). Scott K
