HA: Sync Reindex Execution Mode (Update First then Delete Last Stale Documents without Deleting Everything First with Elasticsearch, aka. Soft Reindex)

Priority

Components

Affects versions

None

Fix versions

Description

Goal

Administrators are provided with an option to perform a reindex in synchronize (aka. "soft") execution mode to update documents in the search index without deleting the existing documents (unless there are stale ones) to provide high-availability in search capabilities throughout the process.

This will come in handy to propagate changes in Indexer logic (ie, new fields), fix issues like orphan documents / stale data / corrupt index / missing results with minimal impact on the live search traffic, or to synchronize the index after restoring it from a snapshot.

Dependencies

  1. Put Mappings (on Demand or at Startup): for new fields (out-of-scope)

  2. Timestamp in Documents (in-scope)

Details

https://liferay.atlassian.net/l/cp/pkr0nRta#SOFT-(SINGLE-MODEL)-REINDEX

100% Done
Loading...

Activity

Show:

Tibor Lipusz June 15, 2023 at 8:02 AM

Sync Reindex WIP Demo Notes/Action Times

  • Indicate in the product (and in the docs) that the sync mode require an initial full reindex to create the mapping for and populate the timestamp field for all documents on upgraded systems.

    • Could we detect when it's actually happened? => Reindex auditing ()?

  • Put mapping for the timestamp field on startup?

  • Stale document detection & reporting: revisit the idea to implement something similar to Jorge's index checker (https://marketplace.liferay.com/p/index-checker)

  • Allow to configure the time range (Conditional Reindex, )?

Completed
Pinned fields
Click on the next to a field label to start pinning.

Details

Assignee

Reporter

Main Core Job

Operate - Administer and Maintain

Feature Flag

On

Due date

Zendesk Support

Created March 7, 2023 at 2:19 AM
Updated November 6, 2024 at 10:20 AM
Resolved July 24, 2023 at 4:43 PM