Updating the Elvis 6 Server index

When users search for an asset, Elvis uses an index to locate where the asset is stored. It is important therefore that the index is always up-to-date.

Available indexes and when to update them

Elvis contains multiple indexes but the only one that you will ever typically update is the 'asset' index.

Note: Updating the asset index is mainly necessary when: Changing custom metadata in the custom-assetinfo.xml file You are prompted by Elvis Server to do so

Other available indexes include:

• config
• Enterprise-updates
• import
• processing
• relation
• security
• settings
• stats
• users
• various
• version

Whenever an index needs to be manually updated, this will be mentioned in a Help Center article.

Updating the index

Note: Updating an index might take a while, depending on the size of your index.

Updating the index is done by using the Index Revision Tool in the Management Console.

To update an index, click its Re-index button.

When the indexing succeeds, the revision number will be incremented and your new configuration will have taken effect. If it fails, you will get an error message and no changes will have been made to the index.

Note: See the comments at the end of this article for more information.

Optimizing the indexing performance

The indexing performance of Elasticsearch can be optimized by disabling its merge throttling during the indexing process. This will speed up the process of indexing significantly. Throttling can be enabled again when the indexing is completed.

Notes: Only set this property if users have no access to the system at the time, as it will have a high impact on performance. These steps can be applied while the re-indexing process is ongoing.

Disabling throttling is done by setting the indices.store.throttle.type option to 'none': 

curl -XPUT 'localhost:9200/_cluster/settings' -d '{ "transient" : { "indices.store.throttle.type": "none" }}'

The throttle setting will reset after a full restart of the cluster as it is transient. To avoid the need of a restart, it can be reset manually to the original value:

curl -XPUT 'localhost:9200/_cluster/settings' -d '{ "transient" : { "indices.store.throttle.type": "merge" }}'

Troubleshooting

Why is the config index always set to '1' after restarting the server?

At each server restart, the config index is regenerated and redistributed across all nodes and will therefore always be '1', even when initially incrementing it by updating the index.

Errors appear in the log files after updating the asset index

The following errors might appear in the log files:

ERROR com.ds.acm.ak - NEW CONFIGURATION IS INCORRECT: Compass settings of the field language were changed. It is incorrect to modify the field.

WARN com.ds.acm.warnings.Server - warning: Detected illegal changes in new asset info

com.ds.acm.al: New configuration is incorrect, see log for details. Please review asset info configuration.

These errors are generated by a safety feature that is in place during the Elvis service startup which compares the full-assetInfo.xml file (created during the last successful startup) with the new-assetInfo.xml file (constructed from the (internal)default-assetinfo file and the possible (custom) configured custom-assetinfo file).

If the compare results in a difference, the above errors are thrown in the Elvis Server log with a description where this difference is found.

This safety feature is in place to protect the system from starting with an incorrect assetinfo file.

A common scenario for this situation to occur is when changes have been made to the configuration of the assetInfo in the custom-assetinfo.xml file. It may also occur after upgrading Elvis to a newer version with an updated default assetinfo.

It is important to be aware if recent changes have been made to your custom-assetInfo.xml:

• Yes, I just updated the configuration of the custom-assetInfo file which resulted in this log message. Please read the log message carefully; it might describe where a possible configuration error occurred.

• No, I just upgraded my Elvis Server version (with a new/changed default assetinfo file) which resulted in this log message.

Solution 1

Update the asset index once more.

Solution 2

Do the following:

Step 1. Stop your Elvis Cluster nodes.

Step 2. On each node, access the Elvis Server/Elvis Hot Data/elvis-data/assetInfo/ folder and remove the following files:

• full_assetinfo.xml
• new_assetinfo.xml

Note: Make sure you do this on every node (including the nodes that act as the Processing Server and job runner).

Step 3. Start the Elvis Cluster.

New files will be created and the errors should not appear anymore.

Document history

• 4 February 2021: Added section 'Optimizing the indexing performance´.