#6962 BatchIndexer not flushing properly during imports

v1.0.0
closed
Dave Brondsema
import (116) sf-2 (994)
General
Tim Van Steenburgh
2015-08-20
2013-12-06
Dave Brondsema
No

During imports, the BatchIndexer isn't flushing all the ming sessions properly, so some artifacts (e.g. last ticket in a trac import) aren't in that add_artifacts task. That artifact then gets picked up later in a subsequent import's BatchIndexer session. So then when that add_artifacts task runs, the ticket is from the previous project but the context is this new project. As a result, artifact shortlinks will get resolved to the current project instead of the first project, and then the "related artifacts" will be wrong for that ticket.

Discussion

  • Dave Brondsema

    Dave Brondsema - 2013-12-06
    • status: in-progress --> code-review
     

  • Dave Brondsema

    Dave Brondsema - 2013-12-06

    allura:db/6962

    I manually tested this behavior by importing tickets for https://sourceforge.net/apps/trac/mcclidar and then finding the add_artifacts task at /nf/admin/task_manager/ and seeing if it had the _id matching db.ticket.find({ticket_num:49, app_config_id: from_url('/p/testit/tickets-mcc-batch2')._id}).next()._id

     

  • Tim Van Steenburgh - 2013-12-09
    • QA: Tim Van Steenburgh
     

  • Tim Van Steenburgh - 2013-12-09
    • status: code-review --> closed
     

Log in to post a comment.