Hi there,
I recently updated Kibana from 4.5 to 5.4.1 and Elasticsearch from 2.3 to 5.4.1. After this some indices start look broken:
in 'Discover' I get '"field" is a required parameter' message and it stays in 'Searching' state even if I use time range so small there are no entries in it;
existing visualizations and dashboards don't show any data and I get the same message as in previous item; when I try to add new visualization, it does not recognize any string/number/timestamp field, only geo_point;
if I open corresponding index pattern, no fields are searchable or aggregatable (except geo_point); refreshing field list does not help;
if I create new index through API, the result is the same as in item 3 (does not matter what fields it contains); I use nuget libraries Elasticsearch.Net and NEST, v5.4.0 both; and yes, I use Keyword/Text attributes.
There are the indices that work correct however. The only difference that I see between them is that those that stopped working contain geo_point field (but this field is the last thing working for them).
If it's the wrong category, please move it to the correct one.
sorry for the delay.
a standard procedure, described in docs: update debian repo definition, stop, apt-get install, start, first for Elasticsearch, than for Kibana. since we have only 1 VM, not a cluster, i thought it's not needed to disable shard allocation.
before updating Elasticsearch i installed migration helper plugin and checked that everything is green, except notification that from/to won't be available for geo_point any more (we don't use from/to anyway).
@Dmytro_Orel I'm sorry to tell you that this is being caused by https://github.com/elastic/elasticsearch/issues/24275 which is unresolved at this time. The only work-around that we have at this moment is to reindex your data, or wait for Kibana 5.5 to be released which won't rely on the _field_stats to determine index patterns.
just a basic reindex like first example in guide, or something more sophisticated?
and by the way, how will it fix issue with creating new index? just to remind:
4. if I create new index through API, the result is the same as in item 3 (does not matter what fields it contains); I use nuget libraries Elasticsearch.Net and NEST, v5.4.0 both; and yes, I use Keyword/Text attributes.
@Dmytro_Orel the basic reindex like the first example in the guide will work.
Regarding item 4, if you create a new index using the API and then created an index pattern in Kibana using the exact index name without wildcard names, what are you seeing?
@Dmytro_Orel Interesting... would you mind creating a basic index using CURL to determine if this is being caused by the client libraries or Elasticsearch itself? Does this happen when creating indices without the geo fields?
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.