#5963 Code snapshot status check changes

v1.0.0
closed
nobody
42cc (433)
General
Cory Johns
2015-08-20
2013-03-13
Cory Johns
No

Since the code snapshots are not going to be stored indefinitely, the status check should use test file existence instead of a database flag. The snapshot should be generated with a temporary file name, and atomically renamed to the final file name when complete. The status check should return 'ready' if the final file name exists, 'busy' if the temporary file name exists, and None (and thus initiate a new build) if neither exist.

Additionally, since the snapshot could take a while to build, the delay between status checks should decay from a short delay to a much longer one, on the order of minutes or tens of minutes, instead of being a fixed 5s.

Related

Tickets: #2840
Tickets: #5963

Discussion

  • Dave Brondsema

    Dave Brondsema - 2013-03-13
    • labels: --> 42cc
     
  • Cory Johns

    Cory Johns - 2013-03-13
    • labels: 42cc -->
    • Description has changed:

    Diff:

    --- old
    +++ new
    @@ -1,3 +1,3 @@
    -Since the code snapshots are not going to be stored indefinitely, the status check needs to be changed to test for the existence of the snapshot file.  Along with that, the file should be generated under a temporary name and then atomically renamed to the final name when done, so that in-progress snapshots don't show as complete (though they do need to flag as busy so another generate task isn't started).
    +Since the code snapshots are not going to be stored indefinitely, the status check should use test file existence instead of a database flag.  The snapshot should be generated with a temporary file name, and atomically renamed to the final file name when complete.  The status check should return `'ready'` if the final file name exists, `'busy'` if the temporary file name exists, and `None` (and thus initiate a new build) if neither exist.
    
     Additionally, since the snapshot could take a while to build, the delay between status checks should decay from a short delay to a much longer one, on the order of minutes or tens of minutes, instead of being a fixed 5s.
    
     
  • Dave Brondsema

    Dave Brondsema - 2013-03-13
    • labels: --> 42cc
     
  • Igor Bondarenko

    Igor Bondarenko - 2013-03-14
    • status: open --> in-progress
     
  • Igor Bondarenko

    Igor Bondarenko - 2013-03-14

    Created #305: [#5963] Code snapshot status check changes (2cp)

     

    Related

    Tickets: #5963

  • Igor Bondarenko

    Igor Bondarenko - 2013-03-26
    • status: in-progress --> code-review
     
  • Igor Bondarenko

    Igor Bondarenko - 2013-03-26

    Closed #305.

    Allura: je/42cc_5963
    ForgeHg: 42cc_5963

     
  • Cory Johns

    Cory Johns - 2013-03-29
    • qa: Cory Johns
     
  • Cory Johns

    Cory Johns - 2013-04-01
    • status: code-review --> closed
    • milestone: forge-backlog --> forge-apr-05