|author||Jack Vanlightly <email@example.com>||Thu Mar 11 23:20:31 2021 -0800|
|committer||Enrico Olivelli <firstname.lastname@example.org>||Fri Mar 12 08:20:50 2021 +0100|
Add fencing to recovery reads to avoid data loss issue Descriptions of the changes in this PR: ### Motivation Adding the fencing flag to recovery reads avoids a data loss scenario as described in [issue 2614](https://github.com/apache/bookkeeper/issues/2614) ### Changes Added the fencing flag to recovery reads. Refactored some mocking unit test code to introduce fencing and allow two writers to share the same bookie state. Added a new unit to verify the fix. You can recreate the initial problem by removing the fencing flag from recovery reads and running the new unit test. Master Issue: #2614 Reviewers: Enrico Olivelli <email@example.com>, Andrey Yegorov <firstname.lastname@example.org> This closes #2616 from Vanlightly/fix-fencing (cherry picked from commit 017307bc67431a7616861ad09927c4e3327633d1) Signed-off-by: Enrico Olivelli <email@example.com>
Apache BookKeeper is a scalable, fault tolerant and low latency storage service optimized for append-only workloads.
It is suitable for being used in following scenarios:
Join us on Slack - This is the most immediate way to connect with Apache BookKeeper committers and contributors.
We feel that a welcoming open community is important and welcome contributions.
See Developer Setup to get your local environment setup.
Make your changes according to our contribution guide.