yahoo-eng-team team mailing list archive

[OpenStack Infra <1367771@xxxxxxxxxxxxxxxxxx>]

Reviewed:  https://review.openstack.org/126856
Committed: https://git.openstack.org/cgit/openstack/glance/commit/?id=76c66343a45cba0068c97d1ad21b46c63977e13d
Submitter: Jenkins
Branch:    proposed/juno

commit 76c66343a45cba0068c97d1ad21b46c63977e13d
Author: Bartosz Fic <bartosz.fic@xxxxxxxxx>
Date:   Tue Oct 7 18:13:51 2014 +0200

    Use ID for namespace generated by DB
    
    In current implementation ID that is used in namespace to
    insert data to DB is generated by built-in function - enumerate.
    This causes problems with loading the metadata when there are already
    namespaces in DB.
    
    This patch removes 'enumerate' and asks for namespace ID
    generated by database.
    
    Closes-Bug: #1367771
    Co-Authored-By: Bartosz Fic <bartosz.fic@xxxxxxxxx>
    Co-Authored-By: Pawel Koniszewski <pawel.koniszewski@xxxxxxxxx>
    (cherry picked from commit 89c04904416270d3c306d430f443a7127c5fc206)
    
    Conflicts:
    	glance/db/sqlalchemy/metadata.py
    
    Change-Id: I235c6310077526cafb898ac007c3601b4d66c9fe


** Changed in: glance
       Status: Fix Committed => Fix Released

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to Glance.
https://bugs.launchpad.net/bugs/1367771

Title:
  glance-manage db load_metadefs will fail if DB is not empty

Status in OpenStack Image Registry and Delivery Service (Glance):
  Fix Released

Bug description:
  To insert data into DB 'glance-manage db load_metadefs' uses IDs for
  namespaces which are generated by built-in function in Python -
  enumerate:

  for namespace_id, json_schema_file in enumerate(json_schema_files,
  start=1):

  For empty database it works fine, but this causes problems when there
  are already metadata namespaces in database. The problem is that when
  there are already metadata definitions in DB then every invoke of
  glance-manage db load_metadefs leads to IntegrityErrors because of
  duplicated IDs.

  There are two approaches to fix this:
  1. Ask for a namespace just after inserting it. Unfortunately in current implementation we need to do one more query.
  2. When this go live - https://review.openstack.org/#/c/120414/ - then we won't need to do another query, because ID is available just after inserting a namespace to DB (namespace.save(session=session)).

To manage notifications about this bug go to:
https://bugs.launchpad.net/glance/+bug/1367771/+subscriptions