Error Message

N/A.

The Hotword Analysis processing uses regular expressions (RegEx) to run pattern matching operations to match the hotwords used in the search to the hotwords in the search hits. It then calculates the various available statistics as displayed in the Review Set.

When the hotwords used in the search have complex patterns, such as combinations of multiple wildcards, multiple spaces, combinations of English and Chinese/Japanese/Korean (CJK or double-wide unicode characters), the RegEx pattern-matching processing can take an extended amount of time when comparing hotwords to messages. This extended processing time is perceived as the Hotword Analysis stalling or stopping.

In the fix versions listed below, the processing now includes a number of retries per message and a timeout per retry. The number of retries and timeouts are configurable, as follows:

1. Open the CA Client whilst logged in as an account having rights to edit the Configuration Settings, such as the Vault Service Account (VSA).

2. Go to Configuration | Settings | CTRL-Click on the 'Configuration Settings' banner to show the Hidden Settings | Analytics Data Collection.

3. The new Settings are:

3.1. Regular expression engine timeout (seconds).
Description: Regular expression engine timeout for matching hotword to the item content for hotword analysis.
Default value: 5
Minimum value: 1
Maximum value: 60

3.2. Temporarily failed item (due to regular expression engine timeout) retry count.
Description: Number of times to retry a temporarily failed item before marking it as permanently failed, if it was failed due to regular expression engine timeout.
Default value: 3
Minimum value: 0
Maximum value: 2147483647

By default, the Hotword Analysis will try an item 3 times with a timeout of 5 seconds before marking it as failing the analysis.

Notes

1. Increasing the timeout and/or number of attempts will extend the overall time taken for Hotword Hit Analysis to complete. If any changes are to be made to these settings, it is recommended to use small differential changes and to test each change before making any more changes. A small differential change would be editing the timeout from 5 seconds to 7 seconds, or increasing the number of attempts from 3 to 5. Large differential changes could cause the processing to be even further extended. After saving any changes to these Settings, acknowledge any prompts to restart background tasks, remoting or services, and then restart the Enterprise Vault Accelerator Manager Service/IIS Admin service as prompted.

2. Whilst the algorithm has been improved, it cannot account for every conceivable complex combination pattern, as this would be an almost infinitely large list. Once the timeouts and retry attempts have been reached, the processing will fail the message and may log Event ID 596 and/or Event ID 597 Information entries in the EV Event Logs on the CA server. For a description of these events and how to obtain more information on the messages and hotwords involved, please see article:

How to interpret the ID values in Hotword Hit Analysis Event ID 596 and Event ID 597 entries

This issue is fixed in the following release(s), available in the Download Center at https://downloads.enterprisevault.com/:
- Enterprise Vault 14.1.2

- Enterprise Vault 14.2.0

Enterprise Vault (EV) Compliance Accelerator (CA) introduced a new feature in version 12.5 and higher called Hotword Analysis. This feature analyses the hotwords that are found within an item. The feature can be enabled or disabled via the Configuration Settings. The Hotword Analysis feature allows a user to: - Filter items on the basis of existing hotwords and hotword sets which help in refining the search results to show only those items that match the hotwords used in the Search.
- Get information about the count of hotwords found in each item.
- View the hotword statistics for the selected item, including the hotword along with the count of the hotword based on the location - whether in the top level message or in its attachments.
- Filter search results on the basis of the status of hotword analysis such as Pending, In Progress, Completed and so on. Under certain circumstances, the analysis may appear to have stalled or stopped and may not show any movement. Restarting the Enterprise Vault Accelerator Manager Service (EVAMS) may cause the analysis to temporarily resume before appearing to stop again or may have no effect.

JIRA: CFT-3909 JIRA: CFT-4056