jackrabbit-oak-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Amit Jain (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (OAK-5908) BlobIdTracker should not resurrect deleted blob ids in a clustered/shared setup after GC
Date Tue, 14 Mar 2017 04:33:41 GMT

     [ https://issues.apache.org/jira/browse/OAK-5908?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Amit Jain updated OAK-5908:
---------------------------
    Description: 
BlobIdTracker can resurrect deleted blob ids from a previous run when running in a clustered
setup by also synchronizing blob references from other cluster/shared nodes which don't have
information about the deleted blob ids.
The effect of this is that when blob gc is executed again it identifies those ids as candidates
and logs a warning when trying to delete them since they had already been deleted in the last
gc execution.

The locally tracked files at each of the instances should be purged after synchronizing with
the datastore.

  was:
BlobIdTracker can resurrect deleted blob ids from a previous run when running in a clustered
setup by also synchronizing blob references from other cluster nodes which don't have information
about the deleted blob ids.
The effect of this is that when blob gc is executed again it identifies those ids as candidates
and logs a warning when trying to delete them since they had already been deleted in the last
gc execution.

The locally tracked files at each of the instances should be purged after synchronizing with
the datastore.


> BlobIdTracker should not resurrect deleted blob ids in a clustered/shared setup after
GC
> ----------------------------------------------------------------------------------------
>
>                 Key: OAK-5908
>                 URL: https://issues.apache.org/jira/browse/OAK-5908
>             Project: Jackrabbit Oak
>          Issue Type: Bug
>          Components: blob
>            Reporter: Amit Jain
>            Assignee: Amit Jain
>              Labels: candidate_oak_1_6
>             Fix For: 1.7.0, 1.8
>
>
> BlobIdTracker can resurrect deleted blob ids from a previous run when running in a clustered
setup by also synchronizing blob references from other cluster/shared nodes which don't have
information about the deleted blob ids.
> The effect of this is that when blob gc is executed again it identifies those ids as
candidates and logs a warning when trying to delete them since they had already been deleted
in the last gc execution.
> The locally tracked files at each of the instances should be purged after synchronizing
with the datastore.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message