If user configure scheduler to run every month and expiry period is one year then, it creates a file with copy of all the expired notification histories (1 year old) and delete them from the Elastic Search database every month.

To configure scheduler for archiving Notification:


  1. Create an NFS Shared folder (in any of the 3 ES clusters or any other server), and run below command in all machines:
    yum install nfs-utils nfs-utils-lib
        create the folder to be shared across cluster in any machine
                    Ex: mkdir /var/unixmen_share/
        sudo chown -R elasticsearch:elasticsearch /var/unixmen_share
        edit /etc/exports in the vm from where above folder will be shared
             /var/unixmen_share/     <Cluster1 IP>(rw,sync,no_root_squash,no_all_squash)
            /var/unixmen_share/      <Cluster 2(rw,sync,no_root_squash,no_all_squash)
        service nfs start
        Go to all cluster machines and run below command
            mount -t nfs <nfs_server>:/var/unixmen_share/ /var/unixmen_share/
  2. All Elastic Search cluster nodes must have access to the shared folder to write data into folder.
  3. Verify group id and userid are same for elastic search across clusters, by invoking the below command
    less /etc/passwd | grep elastic search
  4. Add repo in the eventstore.properties configuration file:
    Stop cluster  
        add below in all cluster nodes  /etc/elasticsearch/config/elasticsearch.yml  
        path.repo: ["/var/unixmen_share/"]  
        start the cluster
  5. Create a repo in cluster:
    PUT http://<<Any  ES Cluster IP>>:9200/_snapshot/<<repo_name>>  
            "type": "fs",
            "settings": {
                "location": "/var/unixmen_share/backup_location",
                "compress": true
  6. Configure repo, time of expiry and scheduler time period in:
    // Valid time units are "ns", "ms", "s", "m", "h".
    POST http://<<EPS_Server>>:<<EPS port>>:9090/eps/refresh
      "doc": {
        "purging-scheduler-TimeInterval": "24h",
        "purgingTimeLimit": "now-1y-0M-0d",
        "esdb-repo": "smarts-repo"
  7. Restart Eventstore.
    Note: If esdb-repo is not configured, then it will delete one year old notifications from esdb no backup will be taken.
  8. Check Snapshots are getting created using below API:
    GET http://<<Any ES Cluster IP>>:9200/_cat/snapshots/ <<repo_name>>
  9. To delete snapshots from repo, run below REST API:
    DELETE http://<<Any_ES_Cluster_IP>>:9200/_snapshot/<<repo_name>>/<<snapshot_name>>