I am using the elasticsearch grails plugin for the first time, and when I set the config bulkIndexOnStartup = true, it indexed several of the domain classes correctly, but for a couple of others it seemed to only index one instance out of many, even though I saw 99 distinct instances in the database.
When I tried to specifically index the instances by calling
elasticSearchService.index( anInstance )
it would just delete the old entry and add a new one, so at the end there was still only one instance in elasticsearch. Is there an "indexAllDistinct" setting somewhere? How does it decide to add a new entry or replace an existing entry?
I also noticed that on the instances that were indexed correctly, the _id fields all had numeric values, but on the incorrectly indexed instance, the _id field was "null".
EDIT: I found that the incorrectly indexed classes have the "id" mapping name set to a different field. How can I tell elasticsearch that the "id" has a different name for these classes?
For grails domain classes that map the default "id" to another name, I had to add a transient "id" for elasticsearch. For example:
class LogEntry {
Integer logId // replaces default id
...
static mapping = {
id name: 'logId', column: 'LOG_ID'
...
}
static transients = ['id'] // for elasticsearch
def getId(){
return logId
}
}
Related
I would like to add new column in my existed table using domain class but it is not happening. if I use create-drop in application.yml file then it works but at the same time I lost my data. I needed to keep the data as well as I needed to add new column by updating domain class property in grails 3.3.0 and SQL Server 2012
package com.alumni
class Student
{
String studentId
String studentName
String age
static constraints = {
}
}
Database is not updating with column address
It's unclear from your code snippet, but it sounds like you're trying to add
String address
Note that columns are made NOT NULL by default, so if data exists in the table, it's attempting to add a non-nullable column, with null data on every row, so the column will fail to create. You need to allow this column to be nullable in the constraints block as well
static constraints = {
address(nullable: true)
}
If you add that, the column should create successfully.
Is it possible to fetch a default value in grails if a column is null? If I were to represent following query via grails domain object then how could I achieve it:
SELECT IFNULL(empsalary,0.00) from Employee;
Domain object:
class Employee{
Integer id,
Float empsalary
static constraints = {
id unique: true, blank:false
empsalary nullable:true
}
}
making empsalary nullable false isn't an option due to existing data
validator on empsalary seems to work when inserting rows but not while data fetch
we can consider writing say getEmpSalary() method on domain and perform check there but there are several other fields we need to do this so trying to avoid massive code changes
If you want a default value to come out of the database without having to code anything into your classes, I suggest you update every row where it is null and set it to 0 in the database. If data is getting inserted from another application and that application is allowing a null value, put a 'DEFAULT 0' on your database column.
Grails also offers an "afterLoad" event which is run when a domain object gets loaded from the database. See the documentation here: http://grails.org/doc/2.3.7/guide/GORM.html.
I think you can do this with HQL:
def salary = Employee.executeQuery('SELECT COALESCE(empsalary, 0.0) FROM Employee')[0]
See this SO Question.
Please try setting Float empsalary = 0.0 in your domain object.
I have a complex searchable configuration for a domain class and its associated domain classes. when I search for about 200 results (max:200) it takes too long to respond.
in the result set I have all fields (simple or association) specified for search in my domain class. I need to return only an id list and ignore other fields of domain class. is it possible? I want to do this for speeding up my search. this id list will be used for querying another no-sql db. it seems that fetching all of the fields is slowing down my search.
I think you can achieve what you want (let the property be searchable but not return it) by setting the property store to no.
For example:
class MyDomain {
String name
String email
static searchable = {
email index:'analyzed', store:'no'
name index:'analyzed'
}
}
In this domain I say that name and email are indexed and analyzed (so they can be searched) but the email property is not being stored, so it will be null when the object is returned. For other properties check: http://grails.org/Searchable+Plugin+-+Mapping+-+Searchable+Property
I have a domain class called FoapRequest. I want one of the properties called "approver" to be a list of integers. Order matters, so I've defined the class as described by http://grails.org/doc/latest/guide/GORM.html#sets,ListsAndMaps as a list:
class FoapRequest {
Integer requester
Integer subject
List approver
static hasMany = [foap:FOAP, newFoap:NewFoap, approver:Integer]
...
Just for clarification, FOAP and NewFoap are two other domain objects.
I need to map this class to a particular table in the Oracle database, so I also specify a static mapping with a join table:
static mapping = {
table 'OUR_SCHEMA.FOAP_REQUEST_TABLE
id column : 'ID', generator:'sequence', params: [sequence:'OUR_SCHEMA.FOAP_REQUEST_SEQ']
requester column : 'REQUESTER'
subject column : 'SUBJECT'
approver indexColumn: [name: "APPROVER_IDX"], generator:'sequence', params: [sequence:'OUR_SCHEMA.APPROVER_SEQ'],
joinTable: [name:'OUR_SCHEMA.APPROVER_TABLE',
key: 'ASSOCIATED_REQUEST',
column: 'APPROVER_PIDM',
type: "integer"
]
However, when I try to create a new instance of the FoapRequest object, I get the following error:
Invalid column type
The console displays the following:
Error 2012-08-01 12:29:31,619 [http-bio-8080-exec-9] ERROR errors.GrailsExceptionResolver - SQLException occurred when processing request: [POST] /FOAPauth/foapRequest/saveFoapRequests - parameters:
I am certain that the issue lies with the jointable. The domain model didn't include the joinTable originally- approver was just an Integer type (I realized too late that I was going to need to track multiple approvers).
Here's the SQL for creating the APPROVERS table:
CREATE TABLE "OUR_SCHEMA"."APPROVER_TABLE"
(
"APPROVER_IDX" NUMBER(*,0) NOT NULL ENABLE,
"ASSOCIATED_REQUEST" NUMBER(*,0) NOT NULL ENABLE,
"APPROVER_PIDM" NUMBER(8),
);
I'd prefer to avoid creating an Approver domain class if at all possible, since all I really need to keep track of are the integer identifiers.
So, after much janking with join tables, I determined that the best way to deal with my needs was to simply create an Approver object in my domain model.
class Approver {
Integer pidm
String approvalDecision
Date lastUpdated
Date dateCreated
static belongsTo = [foap: FOAP]
}
To be honest, I'm not really sure why I was trying so hard to avoid this. Possibly because my DBAs use a version control system for table definitions that I find a hair annoying. :)
Regardless, a simple one-to-many relationship between domain classes met all my needs, no join table required.
For those who are still burning to know, I did manage to get a statically mapped join table working using a Map, which was more appropriate for my needs (though not as appropriate for them as a new domain class, and not nearly as simple).
I ended up doing it in a different domain object- FOAP instead of FoapRequest:
import java.util.Map
class FOAP {
...
Map approvalData
...
static mapping = {
table 'OURSCHEMA.FOAP_TABLE'
id column : 'ID', generator:'jpl.hibernate.util.TriggerAssignedIdentityGenerator'
fund column : 'FUND'
org column : 'ORG'
chartOfAccounts column : 'CHART_OF_ACCOUNTS'
permissionType column: 'PERMISSION_TYPE'
foapRequest column: 'REQUEST_ID'
version column : 'VERSION'
approvalData joinTable: [name:'OURSCHEMA.FOAP_APPROVERS',
key: 'FOAP'
]
}
For the table definition, I used the column names similar to those in my original question.
CREATE TABLE "OUR_SCHEMA"."APPROVER_TABLE"
(
"FOAP" NUMBER(*,0) NOT NULL ENABLE,
"APPROVER_IDX" VARCHAR2(255),
"APPROVER_DLT" NUMBER(8),
);
The IDX column was the map object's key, the DLT column its value. I'd recommend against this approach, for anyone who can avoid it. Creating a new domain object is much simpler.
Let's suppose I define a domain class named Client in several datasources and create a unique constraint on the field "name" :
class Client {
static mapping = {
datasources(['ds1', 'ds2'])
}
String name
static constraints = {
name unique: true
}
}
The unique constraint will be applied across all the datasources, i.e. if I create a Client in the datasource ds1 with the name "client1", then I can't create a client with the same name in the datasource ds2.
I would like the field "name" to be unique for a given datasource, not across all datasources, the same way it is possible to define the field "name" to be unique for the given value(s) of one or several other field(s). In fact, it was the way it worked by default with the Datasources plugin before Grails 2.0. Do you know how to do that ?
If that's how it's working, it's a bug. Please create an issue at http://jira.grails.org/browse/GRAILS