You are viewing a plain text version of this content. The canonical link for it is here.

Posted to gitbox@hive.apache.org by GitBox <gi...@apache.org> on 2022/08/24 01:15:38 UTC

[GitHub] [hive] jhungund opened a new pull request, #3544: HIVE-24933: Replication fails for transactional tables having same na…

jhungund opened a new pull request, #3544:
URL: https://github.com/apache/hive/pull/3544

Replication fails for transactional tables having same name as dropped external tables.

Summary of the change:
While setting up the tasks during the repl-load phase of the replication, delay the access to
table metadata until the task execution. This will avoid inconsistent metadata access during
task creation.

Root Cause Analysis:

Background:
During the incremental load phase of replication, all event logs are processed sequentially.
Multiple tasks are spawned/created during the processing of each event.
All the spawned tasks are subsequently, executed sequantially.

Scenario of the issue:
The issue is seen in the following scenario:
1. An external table(Eg. T1) created, is replicated, to target cluster during earlier replication cycles.
2. This external table is dropped.
3. An new managed table with the same name (T1) is recreated.
4. The subsequent repl-load phase fails.

Root cause:
1. The above mentioned operations (table drop and recreation) are propogated to the target cluster
via event logs during the subsequent incremental phase of replication.
2. We create DDL tasks to drop the external table for drop-table event.
3. We also create new tasks to create new managed tables.
4. Some additional events are logged which create tasks to load the newly created table.
5. During the creation of these load-table tasks, we try to access metadata corresponding to the new table.
During normal scenario of a fresh table creation, the metadata store will not have data correspoding to the new table (yet to be created).
However, in this scenario, the old table still exists and hence, we end up using the metadata corrsesponding to old table(external).
We try to use this metadata to create the load tasks for the new table.
During the exeuciton of these load tasks, which execute after the drop and recreate tasks, we find that the metadata set in the
task context is stale and is inconsistent with the newly created table. Hence, the error.

Fix:
Do not access the table metadata during the task creation phase for table-load.
Instead, access the metadata during the task execution. By that time,
the metadata is updated to the latest state with the previously executed tasks.

Change-Id: I79ed804617dcdadb51f961a933f4023ac0b6f509

<!--
Thanks for sending a pull request! Here are some tips for you:
1. If this is your first time, please read our contributor guidelines: https://cwiki.apache.org/confluence/display/Hive/HowToContribute
2. Ensure that you have created an issue on the Hive project JIRA: https://issues.apache.org/jira/projects/HIVE/summary
3. Ensure you have added or run the appropriate tests for your PR:
4. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP]HIVE-XXXXX: Your PR title ...'.
5. Be sure to keep the PR description updated to reflect all changes.
6. Please write your PR title to summarize what this PR proposes.
7. If possible, provide a concise example to reproduce the issue for a faster review.

-->

### What changes were proposed in this pull request?

### Why are the changes needed?

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?
- Unit-Tested

--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: gitbox-unsubscribe@hive.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org

---------------------------------------------------------------------
To unsubscribe, e-mail: gitbox-unsubscribe@hive.apache.org
For additional commands, e-mail: gitbox-help@hive.apache.org