Logging alarm state and other messages to an elastic back end.

1. Elasticsearch version 8.x OS specific release can be found here:
   https://www.elastic.co/downloads/past-releases#elasticsearch
   The CI/CD pipeline is setup to test with elastic release 8.2.3

bin/elasticsearch

elasticsearch defaults to port 9200, however if elasticsearch is configured to use another port this can be set in

/src/main/resources/alarm_logging_preferences.properties

mvn clean install

or

ant clean dist

1. Run the jar

java -jar target/alarm-logger-<version>.jar -topics MY,ALARM,CONFIGS

The argument for the -topics switch is the comma separated list of alarm configurations you wish to be logged. An alarm configuration is the value specified for the -config switch when starting an alarm server instance. See the README.md file in the alarm-server module.

2. Using spring boot

mvn spring-boot:run

Run with -help to see command line options, including those used to create daily, weekly or monthly indices.

The alarm logger can be configured via command line switches when running the jar, see option -help for details, or via properties documented in here

http://localhost:8080

Will provide information regarding the version of the service along with information about the connection status with the elastic backend.

e.g.

{
    "name": "Alarm logging Service",
    "version": "4.7.4-SNAPSHOT",
    "elastic": {
        "status": "Connected",
        "clusterName": "elasticsearch",
        "clusterUuid": "hwn6nGDwTKSm8vzVTqR9Sw",
        "version": "co.elastic.clients.elasticsearch._types.ElasticsearchVersionInfo@51e5581d"
    }
}

Alarm logs can be retrieved using the alarm logging services query REST API's

curl -X GET 'localhost:8080/search/alarm?pv=*'

The most common aspects for effectively configuring the alarm logger are:

The new elastic indices are created based on the templates which are automatically created when the service is first launched.

By default, the alarm logger will create a new index for each month, named {Alarm topic}_state_yyyy_mm_dd and ..._cmd_.... This supports the removal of older data by simply and efficiently deleting older indices.

Queries can read from all these indices by using a pattern like {Alarm topic}_state_*, so they are abstracted from the periodically structured indices.

In practice, monthly indexing, which is also the default, has been proven most useful, keeping the indices for roughly one year. While finer grained (weekly or even daily) indexing is possible, it will likely require more frequent index cleanup.

Obsolete data should be periodically removed, this can be achieved by deleting the indices from elastic which contain stale data.

One or more indices can be deleted with the following command:

curl -X DELETE 'localhost:9200/accelerator_alarms_state_2019-02-*'

Prepare the release
mvn release:prepare
In this step will ensure there are no uncommitted changes, ensure the versions number are correct, tag the scm, etc. A full list of checks is documented here.

Perform the release
mvn -Darguments="-Dskip-executable-jar" -Pdocs,releases release:perform
Checkout the release tag, build, sign and push the build binaries to sonatype. The docs profile is needed in order to create required javadocs jars.