You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@geode.apache.org by "ASF subversion and git services (Jira)" <ji...@apache.org> on 2021/05/04 18:17:00 UTC
[jira] [Commented] (GEODE-8991) Cached regions are not cleaned up
whenever the connection to the cluster is lost
[ https://issues.apache.org/jira/browse/GEODE-8991?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17339206#comment-17339206 ]
ASF subversion and git services commented on GEODE-8991:
--------------------------------------------------------
Commit b5cc77231ffef370a7df4acea6d3021ecbcabcf4 in geode-native's branch refs/heads/develop from Mario Salazar de Torres
[ https://gitbox.apache.org/repos/asf?p=geode-native.git;h=b5cc772 ]
GEODE-8991: Implement cached region clean up (#757) (#799)
- Whenever subscription redundancy is lost cached regions are supposed
to be cleaned up. This commit implements interest recovery to achieve
precisely that.
- Whenever subscription redundancy is lost all entries with interest
registered are cleaned up.
- Implemented an optimization for regions which have interest
registered to all keys.
- ITs implemented to verify the functionality.
- Adapted RegisterAllWithConsistencyDisabled test to work with the
modified CacheListener mock.
- As noted for RegisterAllWithConsistencyDisabled case, a fail-safe
have been added to avoid a rare race condition.
- Fixed coredump when trying to clear interest registry keys for
regions with caching disabled.
- Fixed exception whenever cleaning up cached region which interest
list is a set of keys and some of them were already destroyed.
- Export binary_semaphore class to make it avaialable for ITs. In the
future it would be nice to have a library with all of the utilities
common to the apache-geode library and tests.
> Cached regions are not cleaned up whenever the connection to the cluster is lost
> --------------------------------------------------------------------------------
>
> Key: GEODE-8991
> URL: https://issues.apache.org/jira/browse/GEODE-8991
> Project: Geode
> Issue Type: Bug
> Components: native client
> Affects Versions: 1.15.0
> Reporter: Mario Salazar de Torres
> Assignee: Mario Salazar de Torres
> Priority: Major
> Labels: pull-request-available
>
> Under a cluster restart scenario it can't be guaranteed that all the entries that existed before, exist after the restart. Hence if geode-native client has any cached region registered it should be cleared after the connection is lost towards the cluster in order to ensure cache consistency.
> This is happenig in the case of the Java client, as described in this part of the documentation: [https://geode.apache.org/docs/guide/12/developing/events/how_client_server_distribution_works.html#how_client_server_distribution_works__section_928BB60066414BEB9FAA7FB3120334A3]
> However, this is not the case for the native client
--
This message was sent by Atlassian Jira
(v8.3.4#803005)