@Labidi, is this the same server we discussed in this thread?
You mention there that the specs are 12 GB RAM and 8 CPU, but here you say 16 GB and 8 CPU. To help you allocate memory optimally can you clarify?
If this is the same server being discussed in both threads it’s apparent Elasticsearch is having some issues, in the other thread you did not mention that the Elasticsearch service was having periods of being unreachable which means most likely that it is crashing repeatedly. What errors are present in the Elasticsearch log?
You can view heap size in /etc/elasticsearch/jvm.options