I have a use-case where I create month based indexes in ElasticSearch. The data in these indexes can be updated (append-only to array type fields) if a document already exists in any month based index or else the document will be created in the current month index.
Can I do this with a single operation (append if exists in any index or else create in latest index)? If not, what is the simplest way of achieving this (using JAVA)?
If you are using JAVA API, try Rest High Level Client. You can search the existing document using GET API and you can send an Update Request to Elasticsearch. While Updating the existing document, make sure that you have added the existing meta-data in the content. Index API will be useful for indexing it for the first time.
Once you are familiar with the concepts, instead of sending Update Request, you can directly send Index Request which will be considered as an update request by the Elasticsearch itself.
Related
I'm new to Elastic search and I'm doing a task in which I need to upload more number of documents to ES. Whenever I upload everytime, I need to specify document id for that document in IndexRequest api. Is there any way in java so that i can insert documents without giving id for it (i.e creating random document id's for my documents).
Please have a look at https://www.elastic.co/guide/en/elasticsearch/client/java-rest/7.13/java-rest-high-document-index.html
In order to have the id autogenerated, just ommit this call:
request.id("1");
This should do the trick for single document operations.
If you need bulk changes, see https://www.elastic.co/guide/en/elasticsearch/client/java-rest/master/java-rest-high-document-bulk.html
In this case, also remove the
.id("1")
method call.
I am using elastic search first time.but i can not finalize which api to use for update.it can be done by update api and also index api.but in performance which one is better?
Update API and Index API are two different things. In index API , you can over-write existing whole documents but then Update API , you can change or edit parts of the documents.
Under the hood , both are marking the original document deleted and creating a new document.
I have cloudant database with some already populated documents in use... I'm using a cloudant java client to fetch data from that. I plan to change the indexes that are used currently. Basically I plan to change over from using createIndex() to https://github.com/cloudant/java-cloudant#cloudant-search. Also would like to change the fields on which the documents are indexed.
Would changing the index impact the underlying data or cause any migration issues with existing data when I start to use the new Index?
It sounds like you want to change from using Cloudant Query to Cloudant Search. This should be straight forward and safe.
Adding a new index will not change or affect the existing data -- the main thing to be careful of is not deleting your old index before you've migrated your code. The easiest way to do this is by using a new design document for your new search indexes:
Create a new design document containing your search index and upload it to Cloudant (https://github.com/cloudant/java-cloudant#creating-a-search-index).
Migrate your app to use the new search index.
(Optionally) remove the design document containing the indexes that you no longer need. Cloudant will then clean up the index files that are no longer needed (https://github.com/cloudant/java-cloudant#comcloudantclientapidatabaseremovedoc-idrev-id).
I included links to the relevant parts of the Java API, but obviously you could do this through the dashboard.
I have a Mongodb database that contains a Poll Collection.
The Poll collection has a number of Poll documents. This could be a large number of documents.
I am using Java Servlet for serving HTTP requests.
How can I implement a feed kind of retrieval mechanism at the server side?
For e.g., In the first request, I want to retrieve 1 to 10, documents, then 11 to 20 and so on...
As there is a scroll in the view, i want to get the data from server and send to client.
Does Mongodb provide a way to do this?
I think what you are looking for is a pagination. You could use the limit and skip methods with your find query.
First request
db.Poll.find().skip(0).limit(10)
Second request
db.Poll.find().skip(10).limit(10)
...
...
Note: You should also be sorting your find with some field.
db.Poll.find().skip(10).limit(10).sort({_id:-1})
For more info on the cursor methods you could look here: http://docs.mongodb.org/manual/reference/method/js-cursor/
I want to store java objects as part of the Solr document.
They don't need to be parsed or searched, only be returned as part of the document.
I can convert them to json or XML and store the text but I prefer something more efficient.
If I could use Java serialization and then add the binary blob to the document it could be ideal.
I'm aware of the option to convert the binary blob with base64 but I was wondering if there is a more efficient way.
I do not share the opinions of the first two answers.
An additional database call can in some scenarios be completely unnecessary, Solr can act as a NoSQL database, too.
It can even use compression for some fields, which affects CPU cost, but saves some cache memory for some kind of binary data.
Take a look at BinaryField and the lazy loading field declarations within your schema.xml.
As you can construct an id in Solr to pass with any document, you can store this object in other way (database for example) and query it as you get the id back from solr.
For example, we're storing web pages in Solr. When we index it, we're creating an id which match the id of a WebPage Object created by the ORM in the database
When a search is performed, we get the id back and load the java object from the database
No need to store it in solr (which has been made to store and index documents)