[v1.0.2.rc2] ES role shows the same on all nodes

[dannyk81]

Just upgraded to the latest RC, however I noticed that all nodes have almost the same values for the ES role labels, only semi-correct ones are ingest node (though they show that they are also client node, which they are not):

elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="true",es_master_node="true",host="elasticin-02",instance="elasticin-02",job="elasticsearch-exporter",name="elasticin-02"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="true",es_master_node="true",host="elasticin-01",instance="elasticin-01",job="elasticsearch-exporter",name="elasticin-01"}

The rest are all wrong:

elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticmaster-03",instance="elasticmaster-03",job="elasticsearch-exporter",name="elasticmaster-03"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticmaster-02",instance="elasticmaster-02",job="elasticsearch-exporter",name="elasticmaster-02"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticmaster-01",instance="elasticmaster-01",job="elasticsearch-exporter",name="elasticmaster-01"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticdata-06",instance="elasticdata-06",job="elasticsearch-exporter",name="elasticdata-06"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticdata-05",instance="elasticdata-05",job="elasticsearch-exporter",name="elasticdata-05"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticdata-04",instance="elasticdata-04",job="elasticsearch-exporter",name="elasticdata-04"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticdata-03",instance="elasticdata-03",job="elasticsearch-exporter",name="elasticdata-03"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticdata-02",instance="elasticdata-02",job="elasticsearch-exporter",name="elasticdata-02"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticdata-01",instance="elasticdata-01",job="elasticsearch-exporter",name="elasticdata-01"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticcli-02",instance="elasticcli-02",job="elasticsearch-exporter",name="elasticcli-02"}
elasticsearch_indices_flush_time_seconds{cluster="EFKcluster",es_client_node="true",es_data_node="true",es_ingest_node="false",es_master_node="true",host="elasticcli-01",instance="elasticcli-01",job="elasticsearch-exporter",name="elasticcli-01"}

I double checked our configuration and the roles are defined correctly.

We are using Elasticsearch 5.5.0

Any ideas? any addtional info you need?

[dominikschulz]

Could you provide the output of the following commands?

• curl http://elasticmaster-01:9200/_nodes/stats
• curl http://elasticdata-01:9200/_nodes/stats
• curl http://elasticcli-01:9200/_nodes/stats

[dannyk81]

Please see attached, though perhaps I should run the query with the nodeId? e.g. curl http://elasticmaster-01:9200/_nodes/elasticmaster-01/stats (otherwise you get metrics of all nodes, unless this is the expected result).

node_stats.zip

[dominikschulz]

I'm sorry, of course the output from one of the nodes would have been enough. My bad!

But this is very helpful in narrowing down the issue. I'm using it to build some test cases to prevent us from having regressions on this in the future and then we'll fix the role detection.

[zwopir]

@dannyk81 the fix was merged. I update the docker image and create a new release ASAP

[dannyk81]

@dominikschulz @zwopir

Thank you!

[haizaar]

Hi guys,

We've moved to ES5 and are suffering from this as well. Can you please release a new version?
I tested master and it solves it indeed.

[dominikschulz]

Thank for testing the fix, we'll try to release a new version soon. @zwopir

[haizaar]

Thanks,
Do you have any concrete ETA? The reason I'm asking is that the last time @zwopir said "ASAP" on Nov 27 and it's still not released :)

[dominikschulz]

We've just released the 1.0.2.