Public bug reported: We meet this problem in our production environment.
When users is deleting an image but the glance-api process is down or restarted suddenly, the image's status will be 'deleted', but the 'deleted' column in the image table is False. This lead an error that the image still can be shown, but its status is 'deleted', and then it can't be deleted anymore either by DELETE API call or db purge. I found that the bug related to the code: https://github.com/openstack/glance/blob/master/glance/db/__init__.py#L205 what glance does; 1. mark the image's status as 'deleted' 2. then delete the image's property and mark the image as deleted=True, deleted_at=xxx. If the glance-api process or the DELETE API call is down between step 1 and 2. The case I mentioned above will occur. I have two idea to solve the problem: 1. add a check mechanism to service initialization. When glance-api process start, it should check the db data, purge the remain ones. 2. combine step 1 and step 2 into one sql transaction to avoid data inconsistency. ** Affects: glance Importance: Undecided Status: New ** Description changed: We meet this problem in our production environment. - When users is deleting a image but the glance-api process is down or + When users is deleting an image but the glance-api process is down or restarted suddenly, the image's status will be 'deleted', but the 'deleted' column in the image table is False. This lead an error that the image still can be shown, but its status is 'deleted', and then it can't be deleted anymore either by DELETE API call or db purge. I found that the bug related to the code: https://github.com/openstack/glance/blob/master/glance/db/__init__.py#L205 what glance does; 1. mark the image's status as 'deleted' 2. then delete the image's property and mark the image as deleted=True, deleted_at=xxx. If the glance-api process or the DELETE API call is down between step 1 and 2. The case I mentioned above will occur. I have two idea to solve the problem: 1. add a check mechanism to service initialization. When glance-api process start, it should check the db data, purge the remain ones. 2. combine step 1 and step 2 into one sql transaction to avoid data inconsistency. -- You received this bug notification because you are a member of Yahoo! Engineering Team, which is subscribed to Glance. https://bugs.launchpad.net/bugs/1755411 Title: image can't be deleted in some case Status in Glance: New Bug description: We meet this problem in our production environment. When users is deleting an image but the glance-api process is down or restarted suddenly, the image's status will be 'deleted', but the 'deleted' column in the image table is False. This lead an error that the image still can be shown, but its status is 'deleted', and then it can't be deleted anymore either by DELETE API call or db purge. I found that the bug related to the code: https://github.com/openstack/glance/blob/master/glance/db/__init__.py#L205 what glance does; 1. mark the image's status as 'deleted' 2. then delete the image's property and mark the image as deleted=True, deleted_at=xxx. If the glance-api process or the DELETE API call is down between step 1 and 2. The case I mentioned above will occur. I have two idea to solve the problem: 1. add a check mechanism to service initialization. When glance-api process start, it should check the db data, purge the remain ones. 2. combine step 1 and step 2 into one sql transaction to avoid data inconsistency. To manage notifications about this bug go to: https://bugs.launchpad.net/glance/+bug/1755411/+subscriptions -- Mailing list: https://launchpad.net/~yahoo-eng-team Post to : yahoo-eng-team@lists.launchpad.net Unsubscribe : https://launchpad.net/~yahoo-eng-team More help : https://help.launchpad.net/ListHelp
