Resque task status not properly updated resulting

[Keith Hughitt]

In some cases a task's status is not being properly updated in Resque, resulting in stale movie requests that do not ever get marked as failed/completed.

For example, after making a couple movie requests, resque-web was used to verify that the tasks were no longer running. Further, the cache revealed that both movies has been created and marked as "complete."

Upon querying Helioqueuer, however, the tasks still appear to be in progress:

statuses = r.zrange('resque:_statuses', 0, 10000)
=> ["14e58d40886d012d31eb003048538da3", "0e7318b0890f012d31eb003048538da3", "3d30bd808911012d31eb003048538da3", "6ad572908914012d31eb003048538da3", "b9d87b808916012d31eb003048538da3", "1e72bbc08917012d31eb003048538da3"]
r.get("6ad572908914012d31eb003048538da3")
=> nil
irb(main):019:0> status
=> #<Resque::Status {"uuid"=>"6ad572908914012d31eb003048538da3", "time"=>1281709473, "status"=>"queued"}>
irb(main):020:0> status.completed?
=> false
irb(main):021:0> status.failed?
=> false
r.get 'estimated_time'
=> "2514"

Estimated time never becomes decremented and the status for each stale item are never set to completed, therefor each request for an updated status from hv.org returns the same response.