erickmelo
commented
10 years ago I performed a search on ElasticSearch for each object of my domain class to check if the object was indexed:
def failures=[]
def allObjects = classe.list()
allObjects.each{
elasticSearchHelper.withElasticSearch { client ->
GetRequest getRequest = new GetRequest(getIndexName(domainClass),getTypeName(domainClass), it.id.toString());
def result = client.get(getRequest).actionGet()
if (!result.isExists()){
failures << it
}
}The result (60,000 objects) was about 9,000 failures. After successive indexing the amount was falling. However if I explicitly index objects that failed in the batch, the result is 0 fault (ie, all the objects were indexed correctly). Apparently the bug is in batch operations.
failures.each{
it.index()
}
dawiemalan
sircelsius
noamt
rorybeling
I have some large collections of objects (more than 50 000 in each collection) indexed by Elastichsearch (via Grails Plugin). The problem is that some of these objects are not being indexed properly by the plugin (Domainclass.index () or bulkIndexOnStartup). After successive runs of the indexing process these objects are being indexed (or after performing an explicit indexing on the object - object.index ()).
No exception is reported during the process. I performed some tests directly on the Elastichsearch Client, searching objects by ID and in fact they are not indexed.
Someone has reported this type of behavior and how to handle it?
I believe it would be useful to have some plugin mechanism for consistency checking to detect possible failures, even using heavy operations, such as checking for each object of a given domain class if it is in ElasticSearch index.