Enterprise Search writes events to database table esEvents.
Events can be written by all parts of the Enterprise Search solution, such as discovery, indexing and searching.
The table provides a way to store events that may be of interest for offline analysis, for example in order to improve performance or to collect statistics.
Currently there is no support to add or report events. In order to access table esEvents it is required to have access to the ES database.
The purpose of the events mechanism is primarily analysis. For diagnostics refer to logging mechanisms, for example as based on database table LogEntries and related tables.
Table esEvents has the following columns.
- Nr. A unique, incrementing number.
- Code. A case insensitive event type code, for example MetadataMapper.TryToEsDocumentProperty.UnknownKey.
- nrDocument. The number of the document as registered in the database and as stored in Elastic Search. The number is set if the event applies to a document. The document number identifies a row in tabel esDocuments, and that row identifies a source in table esSources, such that source of the document is known for the event.
- Ms. Duration measurement in ms, for example the number of ms required to obtain a HTML page from a web site.
- Message. Free format message.
- Created. Event creation date time, as UTC.
Enterprise Search includes an automatic cleanup of old events, preserving and condensing events of specific types.
- ES deletes events older than 30 days
- ES deletes old events in excess of the 100.000 newest events.
ES condenses events of specific types. For a series of events having the same code and having the same message ES deletes the oldest events and preserves the single newest event. This applies to the following event types.
- ContentTypeMapper.TryToFileFormat.UnknownContentType. This is a HTTP content type that is not anticipated and not mapped to a file format. For example application/msword may be anticipated and is mapped to MsWord. In contrast application/vnd.ms-word2006ml may not be anticipated and is not mapped. The document is still indexed. Not being able to map the content type only affects a file format facet at search time.
- EsDocumentHelper.SetProperty.DocLanguage. This is a language string that is not anticipated. For example en, english and en-us may be anticpated and map to English. In contrast en, en may not be anticipated. The document will not have a language and the document is not stored in a silo, which is language specific, unless the silo is configured to accept all documents regardless the language.
- MetadataMapper.TryToEsDocumentProperty.UnknownKey. This is the key of a document metadata element that is not anticipated and hence that is not mapped to a standard ES document field. For example dc:title and title may be anticipated and are mapped to the ES document field doc_title. In contrast pdf:docinfo:title may not be anticipated and the content of this metadata element is lost for document indexing and search.
Collected information can be used for configuration or as product improvement feedback.