#!/bin/bash
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.

{% from 'vars/globals.map.jinja' import GLOBALS %}
{% import_yaml 'elasticsearch/defaults.yaml' as ELASTICDEFAULTS %}
{%- set ELASTICSEARCH_HOST = GLOBALS.node_ip -%}
{%- set RETENTION = salt['pillar.get']('elasticsearch:retention', ELASTICDEFAULTS.elasticsearch.retention, merge=true) -%}

LOG="/opt/so/log/curator/so-curator-cluster-delete.log"
LOG_SIZE_LIMIT=$(/usr/sbin/so-elasticsearch-cluster-space-total {{ RETENTION.retention_pct}})

overlimit() {
         [[ $(/usr/sbin/so-elasticsearch-cluster-space-used) -gt "${LOG_SIZE_LIMIT}" ]]
}

# Check to see if Elasticsearch indices using more disk space than LOG_SIZE_LIMIT
# Closed indices will be deleted first. If we are able to bring disk space under LOG_SIZE_LIMIT, we will break out of the loop.
while overlimit; do
  # If we can't query Elasticsearch, then immediately return false.
  /usr/sbin/so-elasticsearch-query _cat/indices?h=index,status > /dev/null 2>&1
  [ $? -eq 1 ] && echo "$(date) - Could not query Elasticsearch." >> ${LOG} && exit
  # We iterate through the closed and open indices
  CLOSED_INDICES=$(/usr/sbin/so-elasticsearch-query _cat/indices?h=index,status | grep 'close$' | awk '{print $1}' | grep -v "so-case" | grep -E "(logstash-|so-|.ds-logs-)" | sort -t- -k3)
  OPEN_INDICES=$(/usr/sbin/so-elasticsearch-query _cat/indices?h=index,status | grep 'open$' | awk '{print $1}' | grep -v "so-case" | grep -E "(logstash-|so-|.ds-logs-)" | sort -t- -k3)
	for INDEX in ${CLOSED_INDICES} ${OPEN_INDICES}; do
	  # Now that we've sorted the indices from oldest to newest, we need to check each index to see if it is assigned as the current write index for a data stream
    # To do so, we need to identify to which data stream this index is associated
    # We extract the data stream name using the pattern below
    DATASTREAM_PATTERN="logs-[a-zA-Z_.]+-[a-zA-Z_.]+"
    DATASTREAM=$(echo "${INDEX}" | grep -oE "$DATASTREAM_PATTERN")
    # We look up the data stream, and determine the write index. If there is only one backing index, we delete the entire data stream
	  BACKING_INDICES=$(/usr/sbin/so-elasticsearch-query _data_stream/${DATASTREAM} | jq -r '.data_streams[0].indices | length')
	  if [ "$BACKING_INDICES" -gt 1 ]; then
      CURRENT_WRITE_INDEX=$(/usr/sbin/so-elasticsearch-query _data_stream/$DATASTREAM | jq -r .data_streams[0].indices[-1].index_name)
	    # We make sure we are not trying to delete a write index
      if [ "${INDEX}" != "${CURRENT_WRITE_INDEX}" ]; then
        # This should not be a write index, so we should be allowed to delete it
        printf "\n$(date) - Used disk space exceeds LOG_SIZE_LIMIT (${LOG_SIZE_LIMIT} GB) - Deleting ${INDEX} index...\n" >> ${LOG}
        /usr/sbin/so-elasticsearch-query ${INDEX} -XDELETE >> ${LOG} 2>&1
      fi
	  else
      # We delete the entire data stream, since there is only one backing index
	    printf "\n$(date) - Used disk space exceeds LOG_SIZE_LIMIT (${LOG_SIZE_LIMIT} GB) - Deleting ${DATASTREAM} data stream...\n" >> ${LOG}
      /usr/sbin/so-elasticsearch-query _data_stream/${DATASTREAM} -XDELETE >> ${LOG} 2>&1
	  fi
    if ! overlimit; then
      exit
    fi
  done
done