Reindexing Content

The search index is kept in sync with CMS content whenever content is created, updated or deleted via the CMS. There are some situations such as direct MySQL database alterations and code changes that will mean the search index has to be repopulated.

When making changes the data or analyzers that are stored in the search index, you must re-index the content so all results and search operations are using the new data specification. Re-indexing is done via CLI commands, as they are potentially long running, high memory usage tasks.

The elasticpress CLI command is used for all index management operations.

As these tasks are long-running commands, it is recommended to run the index on individual sites instead of using the --network-wide flag. When using the --network-wide flag on sufficiently large networks, the CLI process would run out of memory and be killed by the operating system. To better manage memory, it is recommended to use an xargs pipeline to process each site. To re-sychronize the index for all sites:

wp site list --field=url | xargs -I % wp elasticpress index --url=%

wp site list --field=url will print each site URL on a new line and ingested by xargs to be processed individually.

Note: During the indexing process, the site will not use the Elasticsearch index for searches/queries, so some features like faceting or autosuggest will not work as expected until the process is finished.

To re-sychronise the index and update mappings (needed when the mapping has been modified via filters, site language has been changed or a user dictionary has been added and not reindexed successfully):

wp site list --field=url | xargs -I % wp elasticpress index --url=% --setup && wp elasticpress recreate-network-alias

CLI Recommendations

When using CLI to perform re-indexes, it's recommended to set the --per-page argument to a figure around 200. By default, this figure is set to 350 which will often cause service timeouts. For example:

wp site list --field=url | xargs -I % wp elasticpress index --url=% --per-page=200