2149: Upgrade from 1.5.5 to 1.6RC1 breaks the review count

marc.s*******@gmai***** (Google Code) (Is this you? Claim this profile.)
chipx86
 chipx86
July 10, 2011 
What version are you running?
1.6RC1


What steps will reproduce the problem?
I just upgraded to version 1.6RC1 from 1.5.5 and the reviews count has been reset to 0. Now, whenever we close a review (that was created in 1.5.5), the count goes below 0.


What operating system are you using? What browser?
ReviewBoard is deployed on Debian. Using Chrome on the client side
david
#1 david
  • +Milestone-Release1.6
chipx86
#2 chipx86 
It concerns me that you're hitting this, because it was tested to not happen initially, and we're close to release. Can you tell me what database you're using?
  • +NeedInfo
chipx86
#3 chipx86 
Also, which reviews count(s) specifically? Are you seeing these for groups? Incoming? Outgoing? Knowing exactly which ones will help a lot.
#4 marc.s*******@gmai***** (Google Code) (Is this you? Claim this profile.) 
We are using MySQL as the DB.

Looks like all the count are wrong - see attach.
chipx86
#5 chipx86 
Mind trying something?

From the shell, run:

    $ rb-site manage /path/to/your/site shell
    >>> from reviewboard.accounts.models import LocalSiteProfile
    >>> from reviewboard.reviews.models import Group
    >>> LocalSiteProfile.objects.update(direct_incoming_request_count=None, total_incoming_request_count=None, pending_outgoing_request_count=None, total_outgoing_request_count=None, starred_public_request_count=None)
    >>> Group.objects.update(incoming_request_count=None)

Then reload the dashboard and see if the counts appear correct.
#6 marc.s*******@gmai***** (Google Code) (Is this you? Claim this profile.) 
It did work, now all the counts are OK.

Thanks!
chipx86
#7 chipx86 
Okay. So the initial value should have been NULL for each of those (which is what the above script resets to). I don't know how you ended up with 0, and I hope that that's not going to hit everyone. However, I've updated rb-site upgrade to reset all these on upgrade (doing the equivalent of the above script) so the end result should be correct.

Closing, but we may want to revisit this if we see other weirdness.
  • -NeedInfo
    +Fixed
  • +chipx86