Closed midN closed 7 years ago
@midN Graylog runs an embedded Elasticsearch client node for which the file in /etc/graylog/server/elasticsearch.yml
is being used.
The configuration file in /etc/elasticsearch/elasticsearch.yml
is being used by the standalone Elasticsearch node which also stores data (in contrast to the embedded ES node in Graylog which doesn't store any data).
Provided variables in README coming from ES role/playbook are invalid.
These variables are used by Elasticsearch role to build
elasticsearch.yml
file, however Graylog playbook creates its own file in its own defined directory, why is that?As a result i cannot change
es_heap_size
or anything the way it's provided in README as it will not be put in main config created by Graylog playbook. Also there are now 2elasticsearch.yml
configs created, one by Graylog in/etc/graylog/server/elasticsearch.yml
and one in/etc/elasticsearch/<NODENAME>/elasticsearch.yml
.So basically This Task is quite useless and confusing.
In order to make it work properly i had to:
graylog_elasticsearch_config_file
to path used by Elasticsearch playbook.es_user
andes_group
tograylog
in order to have graylog access elasticsearch config.I propose adding
graylog_elasticsearch_config_file
,es_group
andes_user
to README with explanation on how to use them, as well as removing or changing. Otherwise it's impossible to understand/conf it properly due to README in examples having ES config options that in will not be used in the end-result with same README example.