Query Service

The Watson Explorer Engine query service uses local configuration information to identify the collection service that is associated with a given search collection, and then contacts that collection service to locate the indexer service that is associated with that search collection. The query service then forwards the query to that indexer service, and retrieves and returns the results of the query.

Once started, the query-service will automatically restart itself if it crashes. The query-service maintains a set of general-purpose threads to process the web requests for search results or cached pages. The service attempts to maintain at least Minimum general idle threads and at most Maximum general idle threads ready for handling new requests. Similarly, for each collection served, the service attempts to keep Minimum per collection idle threads and at most Maximum per collection idle threads ready for new requests. The defaults should normally be acceptable.

A system that accepts queries and processes them is vulnerable to becoming overloaded. In such a state, queries are arriving faster than the system can process them, causing the system to start doing more concurrent work and never finishing any of it. To protect against this, you can limit the number of concurrent queries to be processed. Queries will not be rejected, but will be delayed until resources are available to handle them.

Important: Instructing the query service to shut down (via the API or by using the Watson Explorer Engine administration tool's Management -> Search Engine tab) when that service is under heavy query load may result in that service becoming unresponsive rather than terminating. In this case, you must terminate the service at the operating system level. However, there is normally no reason to manually shut down the query service because that is usually only done when shutting down Watson Explorer Engine. To shut down Watson Explorer Engine and terminate all Watson Explorer Engine-related processes, use the velocity-shutdown command.

To monitor the health of the query-service, you may elect to have it send an email message when it starts running. This will let you know if the service is crashing for some reason. You can enter a set of email addresses in the Emails option and you can customize the From: field in the Emails from field.

The query service accepts HTTP connections on a specified port. It does not accept HTTPS connections, so allowing remote access to the query service makes all of your data easily accessible. If you have any sensitive data, you should only allow connections from the host on which the query service is running (through its loopback address, 127.0.0.1) and also allow access from the addresses of any front-end machines that you are using to submit queries. By default, the loopback address is the only IP address that can access the query-service. You can allow other hosts, specified by IP address or a wildcard (such as such as 192.168.0.*) by entering approved IP addresses or wildcards that match them in the Allowed IPs field on the Management > Services > Search Engine > Options screen in the Watson Explorer Engine administration tool.

Finally, the My URL option lets you change the URL used for search requests and cache requests. This is useful if you have multiple names for the computer, and the Admin interface selects the wrong one, or if you have firewalls that are doing port forwarding. Enter the address as http://host:port/search (exactly). The query-service is very sensitive to the format of this option.