The indexer service saves the indexes for the fields that are selected and makes them searchable. Indexing fields consumes disk space and processing power.

This section lists the limitations of the indexer service, and provides instructions for configuring indexing for logstores.

Limitations:

  • Messages are tokenized based on the specified separator characters. Only the first 512 tokens are indexed in a message, the rest are ignored. This limitation does not affect other static fields (PROGRAM, HOST, and so on) or name-value pairs added by the pattern database or values coming from the SDATA part of incoming messages.

  • Whitespace characters (space, tabulator and so on) are always treated as delimiters.

  • Tokens that are shorter than 2 characters are not indexed.

  • Tokens are truncated to 59 characters. Therefore, tokens with at least 59 characters long common prefix will be handled as identical ones.

  • When indexing name-value pairs, the 59 characters limitation is applied to this format: "<name-of-nvpair>=<value-of-nvpair>". Do not use long name parts, in order to avoid the premature truncation of the value part.

  • The shortest timeframe for searching and creating statistics is 1 second. Smaller interval cannot be used.

  • The order of the tokens in a message is not preserved. Therefore, if one message contains 'first_token second_token' and another message contains 'second_token first_token' search expressions such as 'first_token second_token' will find both messages.

To configure the indexer service

  1. Navigate to Log > Logspaces and select the logstore to index.

  2. To enable automatic indexing of the logstore files, select the Enable option of the Indexer field.

  3. To limit the number of hits when searching in the logstore, enter the maximum number of search result hits in the Maximum number of search results field.

    To disable the limit, enter 0.

  4. Enter the maximum amount of memory the indexer can use for the current logspace in the Memory limit field.

    		 Caution:

    Hazard of data loss. Increasing the Memory limit option too high (1280 MB) can cause message loss and degraded performance. The exact values that can cause problems depend on your configuration and environment.

    Make sure that the sum of the memory of all indexed logspaces is smaller than the available memory in your syslog-ng Store Box (SSB) appliance.

  5. Configure the fields to be indexed in the Indexed fields.

    NOTE: At least one field must be selected.

    The following fields can be indexed: Facility, Priority, Program, Pid, Host, Tags, Name/value pairs, Message.

    For the Name/value pairs field, select All to index all Name/value fields or enter the names to be indexed in the Only with the name field as comma-separated names.

    If the indexing of the Message field is enabled, the current Delimiters are displayed. By default, the indexer uses the following delimiter characters to separate the message into words (tokens): & : ~ ? ! [ ] = , ; ( ) ' ".

    If your messages contain segments that include one of these delimiters, and you want to search for these segments as a whole, remove the delimiter from the list. For example, if your log messages contain MAC addresses, and you want to be able to search for messages that contain a particular MAC address, delete the colon (:) character from the list of delimiters. Otherwise, the indexer will separate the MAC address into several tokens.

    NOTE: It is not possible to search for the whitespace ( ) character in the MESSAGE part of the log message, since it is a hard-coded delimiter character.