Tranquility not sending data to Druid - druid

I am evaluating Druid for my use case which ingest csv data through tranquility in real time. Following is the server configuration:-
{
"dataSources" : {
"audience" : {
"spec" : {
"dataSchema" : {
"dataSource" : "audience",
"parser" : {
"type" : "string",
"parseSpec":{
"format" : "csv",
"timestampSpec" : {
"column" : "timestamp"
},
"columns" : ["timestamp","partner_id","event_id","product_id","device_id","count"],
"dimensionsSpec" : {
"dimensions" : ["partner_id","event_id","product_id","device_id"]
}
}
},
"metricsSpec" : [{ "type" : "longSum", "name" : total, "fieldName" : "count" }],
"granularitySpec" : {
"segmentGranularity" : "HOUR",
"queryGranularity" : "HOUR",
"intervals" : [ "2013-08-31/2013-09-01" ]
}
},
"ioConfig" : {
"type" : "realtime"
},
"tuningConfig" : {
"type" : "realtime",
"maxRowsInMemory" : "100000",
"intermediatePersistPeriod" : "PT10M",
"windowPeriod" : "PT10M"
}
},
"properties" : {
"task.partitions" : "1",
"task.replicants" : "1"
}
}
},
"properties" : {
"zookeeper.connect" : "localhost",
"druid.discovery.curator.path" : "/druid/discovery",
"druid.selectors.indexing.serviceName" : "druid/overlord",
"http.port" : "8200",
"http.threads" : "8"
}
}
data is generated randomly by a python script as:-
1471336991,1,960,136,3ZLA7,1
1471336991,1,369,367,8MP2B,1
1471336991,2,544,550,C9ZG8,1
1471336991,1,135,394,XFX31,1
1471336991,2,590,552,VXMTL,1
1471336991,1,493,615,0C2HR,1
1471336991,2,435,710,HKYP0,1
1471336991,1,394,483,V2HP9,1
1471336991,2,441,376,J1LYO,1
Following commands submits the data and returns {"result":{"received":1000,"sent":0}}
python createData.py |curl -XPOST -H'Content-Type: text/plain' --data-binary #- http://localhost:8200/v1/post/audience.

Finally able to solve the problem. Actually I was sending time to Druid in Epoch time format, but it expect ISO-8601 format. In python one can easily get so by :-
datetime.datetime.utcnow().isoformat()

Druid supports multiple time formats which can be specified in the "timestampSpec" property.
The Druid documentation lists the following timestamp formats: "iso, millis, posix, auto or any Joda time format."
For example, to send time in milliseconds:
"timestampSpec" : {
"column" : "timestamp",
"format" : "millis"
}

A couple of things
Use ISO 8601 Datetime format
Make sure the written timestamp is within +/- 10Mins of the current hour

Related

NiFi: Converting a Nested JSON File into CSV

I'm new to Nifi and in my case have to consume a JSON topic from Kafka
and would like to convert it into a CSV file where I need to select only few scalar and some nested fields.
I need to do the following things:
Consume topic - Done
Json to CSV
Include header in the CSV file
Merge into single file (if its split)
Give a proper filename with date
Following the below link:
https://community.hortonworks.com/articles/64069/converting-a-large-json-file-into-csv.html
But not sure if this is the right approach and also don't know how to make a single file.
Im using NiFi (1.8) and schema is stored in Confluent Schema Registry
Json Sample:
{
"type" : "record",
"name" : "Customer",
"namespace" : "namespace1"
"fields" : [ {
"name" : "header",
"type" : {
"type" : "record",
"name" : "CustomerDetails",
"namespace" : "namespace1"
"fields" : [ {
"name" : "Id",
"type" : "string"
}, {
"name" : "name",
"type" : "string"
}, {
"name" : "age",
"type" : [ "null", "int" ],
"default" : null
}, {
"name" : "comm",
"type" : [ "null", "int" ],
"default" : null
} ]
},
"doc" : ""
}, {
"name" : "data",
"type" : {
"type" : "record",
"name" : "CustomerData"
"fields" : [ {
"name" : "tags",
"type" : {
"type" : "map",
"values" : "string"
}
}, {
"name" : "data",
"type" : [ "null", "bytes" ]
"default" : null
} ]
},
"doc" : ""
} ]
}
Please guide me.
Try ConvertRecord with a JsonTreeReader and CSVRecordSetWriter. The writer can be configured to include the header line, and you won't need to split/merge. UpdateAttribute can be used to set the filename attribute which is the associated filename for the flow file (used by PutFile, e.g.).
If ConvertRecord doesn't give you the output you want, can you please elaborate on the difference between what it gives you and what you expect?

Upload Data to druid Incrementally

I need to upload data to an existing model. This has to be done on daily basis. I guess some changes needs to be done in the index file and i am not able to figure out. I tried pushing the data with the same model name but the parent data was removed.
Any help would be appreciated.
Here is the ingestion json file :
{
"type" : "index",
"spec" : {
"dataSchema" : {
"dataSource" : "mksales",
"parser" : {
"type" : "string",
"parseSpec" : {
"format" : "json",
"dimensionsSpec" : {
"dimensions" : ["Address",
"City",
"Contract Name",
"Contract Sub Type",
"Contract Type",
"Customer Name",
"Domain",
"Nation",
"Contract Start End Date",
"Zip",
"Sales Rep Name"
]
},
"timestampSpec" : {
"format" : "auto",
"column" : "time"
}
}
},
"metricsSpec" : [
{ "type" : "count", "name" : "count", "type" : "count" },
{"name" : "Price","type" : "doubleSum","fieldName" : "Price"},
{"name" : "Sales","type" : "doubleSum","fieldName" : "Sales"},
{"name" : "Units","type" : "longSum","fieldName" : "Units"}],
"granularitySpec" : {
"type" : "uniform",
"segmentGranularity" : "day",
"queryGranularity" : "none",
"intervals" : ["2000-12-01T00:00:00Z/2030-06-30T00:00:00Z"],
"rollup" : true
}
},
"ioConfig" : {
"type" : "index",
"firehose" : {
"type" : "local",
"baseDir" : "mksales/",
"filter" : "mksales.json"
},
"appendToExisting" : false
},
"tuningConfig" : {
"type" : "index",
"targetPartitionSize" : 10000000,
"maxRowsInMemory" : 40000,
"forceExtendableShardSpecs" : true
}
}
}
There are 2 ways using which you can append/update the data to an existing segment.
Reindexing and Delta Ingestion
You need to reindex your data every time new data comes in a particular segment.(In your case its day) For the reindexing you need to give all the files having data for that day.
For Delta Ingestion you need to use inputSpec type="multi"
You can refer the documentation link for more details - http://druid.io/docs/latest/ingestion/update-existing-data.html

MongoDB insert operation using Groovy script for Jmeter

i am trying to load test MongoDB using Jmeter, i am using JSR223Sampler using Groovy, i am able to connect but for some reason insert part is not working
i need to insert below :
"cart" : {
"schema" : "http://dell.com/dcp/schemas/cart/3.0.0#",
"_id" : "s5ChQonvAUGKM6s2Yq8Z31",
"createdOn" : {
"DateTime" : ISODate("2018-03-07T06:54:01.242Z"),
"Ticks" : NumberLong(636560222412422269),
"Offset" : 330
},
"lastModifiedOn" : {
"DateTime" : ISODate("2018-03-07T06:54:01.245Z"),
"Ticks" : NumberLong(636560222412452266),
"Offset" : 330
},
"expiresOn" : {
"DateTime" : ISODate("2019-04-10T08:21:43.984Z"),
"Ticks" : NumberLong(636904813039840000),
"Offset" : 0
},
"commerceContext" : {
"region" : "us",
"country" : "US",
"language" : "en",
"currency" : "USD",
"segment" : "bsd",
"customerSet" : "rc1005388",
"accessGroup" : "DSA",
"companyNumber" : "08",
"businessUnitId" : "11",
"classCode" : "string",
"sourceApplicationName" : "OLRGCOMM"
},
"items" : [],
"shipments" : [],
"price" : {
"couponCodes" : []
},
"references" : [
{
"referenceId" : "8TOOOrdEJUeiGPTqWA226Q",
"referenceType" : "New Cart",
"referencedOn" : {
"DateTime" : ISODate("2018-03-07T06:54:01.239Z"),
"Ticks" : NumberLong(636560222412392112),
"Offset" : 330
},
"referenceCreatedBy" : "DCQO",
"targetSystem" : "DSP",
"target" : "string"
}
],
"validation" : {},
"properties" : {}
}
})
First of all you need to get MongoDB connection from the MongoDB Source Config, it can be done as follows:
import com.mongodb.DB;
import org.apache.jmeter.protocol.mongodb.config.MongoDBHolder;
DB db = MongoDBHolder.getDBFromSource("mongodb source name", "database name");
Next you just need to call DBCollection.insert() function like:
db.getCollection('your collection name').insert(your DBObject payload here)
More information: How to Load Test MongoDB with JMeter

Druid:how to add a numeric data to metric without aggregation function

The scenario is i want to setup a stock quote server and save the quote data into druid.
my requirement is to get the latest price of all the stock by a query.
But i notice that the query interface of druid such as time series only work on metrics filed ,not the dimension fields.
so i consider to make the price filed one of the metrics,but no need to aggregated.
how can i do it?
Any suggestions?
here is my tranquility config file.
{
"dataSources" : {
"stock-index-topic" : {
"spec" : {
"dataSchema" : {
"dataSource" : "stock-index-topic",
"parser" : {
"type" : "string",
"parseSpec" : {
"timestampSpec" : {
"column" : "timestamp",
"format" : "auto"
},
"dimensionsSpec" : {
"dimensions" : ["code","name","acronym","market","tradeVolume","totalValueTraded","preClosePx","openPrice","highPrice","lowPrice","latestPrice","closePx"],
"dimensionExclusions" : [
"timestamp",
"value"
]
},
"format" : "json"
}
},
"granularitySpec" : {
"type" : "uniform",
"segmentGranularity" : "HOUR",
"queryGranularity" : "SECOND",
},
"metricsSpec" : [
{
"name" : "firstPrice",
"type" : "doubleFirst",
"fieldName" : "tradePrice"
},{
"name" : "lastPrice",
"type" : "doubleLast",
"fieldName" : "tradePrice"
}, {
"name" : "minPrice",
"type" : "doubleMin",
"fieldName" : "tradePrice"
}, {
"name" : "maxPrice",
"type" : "doubleMax",
"fieldName" : "tradePrice"
}
]
},
"ioConfig" : {
"type" : "realtime"
},
"tuningConfig" : {
"type" : "realtime",
"maxRowsInMemory" : "100000",
"intermediatePersistPeriod" : "PT10M",
"windowPeriod" : "PT10M"
}
},
"properties" : {
"task.partitions" : "1",
"task.replicants" : "1",
"topicPattern" : "stock-index-topic"
}
}
},
"properties" : {
"zookeeper.connect" : "localhost:2181",
"druid.discovery.curator.path" : "/druid/discovery",
"druid.selectors.indexing.serviceName" : "druid/overlord",
"commit.periodMillis" : "15000",
"consumer.numThreads" : "2",
"kafka.zookeeper.connect" : "localhost:2181",
"kafka.group.id" : "tranquility-kafka"
}
}
I think you should make [latest_price] as new numeric dimension, it would be much better from performance and querying standpoint considering how druid works.
Metrics and meant to perform aggregation functions as core so won't be helpful in your use case.

How to insert data into druid via tranquility

By following tutorial at http://druid.io/docs/latest/tutorials/tutorial-loading-streaming-data.html , I was able to insert data into druid via Kafka console
Kafka console
The spec file looks as following
examples/indexing/wikipedia.spec
[
{
"dataSchema" : {
"dataSource" : "wikipedia",
"parser" : {
"type" : "string",
"parseSpec" : {
"format" : "json",
"timestampSpec" : {
"column" : "timestamp",
"format" : "auto"
},
"dimensionsSpec" : {
"dimensions": ["page","language","user","unpatrolled","newPage","robot","anonymous","namespace","continent","country","region","city"],
"dimensionExclusions" : [],
"spatialDimensions" : []
}
}
},
"metricsSpec" : [{
"type" : "count",
"name" : "count"
}, {
"type" : "doubleSum",
"name" : "added",
"fieldName" : "added"
}, {
"type" : "doubleSum",
"name" : "deleted",
"fieldName" : "deleted"
}, {
"type" : "doubleSum",
"name" : "delta",
"fieldName" : "delta"
}],
"granularitySpec" : {
"type" : "uniform",
"segmentGranularity" : "DAY",
"queryGranularity" : "NONE"
}
},
"ioConfig" : {
"type" : "realtime",
"firehose": {
"type": "kafka-0.8",
"consumerProps": {
"zookeeper.connect": "localhost:2181",
"zookeeper.connection.timeout.ms" : "15000",
"zookeeper.session.timeout.ms" : "15000",
"zookeeper.sync.time.ms" : "5000",
"group.id": "druid-example",
"fetch.message.max.bytes" : "1048586",
"auto.offset.reset": "largest",
"auto.commit.enable": "false"
},
"feed": "wikipedia"
},
"plumber": {
"type": "realtime"
}
},
"tuningConfig": {
"type" : "realtime",
"maxRowsInMemory": 500000,
"intermediatePersistPeriod": "PT10m",
"windowPeriod": "PT10m",
"basePersistDirectory": "\/tmp\/realtime\/basePersist",
"rejectionPolicy": {
"type": "messageTime"
}
}
}
]
I start realtime via
java -Xmx512m -Duser.timezone=UTC -Dfile.encoding=UTF-8 -Ddruid.realtime.specFile=examples/indexing/wikipedia.spec -classpath config/_common:config/realtime:lib/* io.druid.cli.Main server realtime
In Kafka console, I paste and enter the following
{"timestamp": "2013-08-10T01:02:33Z", "page": "Good Bye", "language" : "en", "user" : "catty", "unpatrolled" : "true", "newPage" : "true", "robot": "false", "anonymous": "false", "namespace":"article", "continent":"North America", "country":"United States", "region":"Bay Area", "city":"San Francisco", "added": 57, "deleted": 200, "delta": -143}
Then I tend to perform query by creating select.json and run curl -X POST 'http://localhost:8084/druid/v2/?pretty' -H 'content-type: application/json' -d #select.json
select.json
{
"queryType": "select",
"dataSource": "wikipedia",
"dimensions":[],
"metrics":[],
"granularity": "all",
"intervals": [
"2000-01-01/2020-01-02"
],
"filter" : {"type":"and",
"fields" : [
{ "type": "selector", "dimension": "user", "value": "catty" }
]
},
"pagingSpec":{"pagingIdentifiers": {}, "threshold":500}
}
I was able to get the following result.
[ {
"timestamp" : "2013-08-10T01:02:33.000Z",
"result" : {
"pagingIdentifiers" : {
"wikipedia_2013-08-10T00:00:00.000Z_2013-08-11T00:00:00.000Z_2013-08-10T00:00:00.000Z" : 0
},
"events" : [ {
"segmentId" : "wikipedia_2013-08-10T00:00:00.000Z_2013-08-11T00:00:00.000Z_2013-08-10T00:00:00.000Z",
"offset" : 0,
"event" : {
"timestamp" : "2013-08-10T01:02:33.000Z",
"continent" : "North America",
"robot" : "false",
"country" : "United States",
"city" : "San Francisco",
"newPage" : "true",
"unpatrolled" : "true",
"namespace" : "article",
"anonymous" : "false",
"language" : "en",
"page" : "Good Bye",
"region" : "Bay Area",
"user" : "catty",
"deleted" : 200.0,
"added" : 57.0,
"count" : 1,
"delta" : -143.0
}
} ]
}
} ]
It seem that I had setup Druid correctly.
Now, I would like to insert data via HTTP endpoint. According to How realtime data input to Druid?, it seems like recommended way is to use tranquility
tranquility
I have indexing service started via
java -Xmx2g -Duser.timezone=UTC -Dfile.encoding=UTF-8 -classpath config/_common:config/overlord:lib/*: io.druid.cli.Main server overlord
conf/server.json looks like
{
"dataSources" : [
{
"spec" : {
"dataSchema" : {
"dataSource" : "wikipedia",
"parser" : {
"type" : "string",
"parseSpec" : {
"format" : "json",
"timestampSpec" : {
"column" : "timestamp",
"format" : "auto"
},
"dimensionsSpec" : {
"dimensions": ["page","language","user","unpatrolled","newPage","robot","anonymous","namespace","continent","country","region","city"],
"dimensionExclusions" : [],
"spatialDimensions" : []
}
}
},
"metricsSpec" : [{
"type" : "count",
"name" : "count"
}, {
"type" : "doubleSum",
"name" : "added",
"fieldName" : "added"
}, {
"type" : "doubleSum",
"name" : "deleted",
"fieldName" : "deleted"
}, {
"type" : "doubleSum",
"name" : "delta",
"fieldName" : "delta"
}],
"granularitySpec" : {
"type" : "uniform",
"segmentGranularity" : "DAY",
"queryGranularity" : "NONE"
}
},
"tuningConfig" : {
"windowPeriod" : "PT10M",
"type" : "realtime",
"intermediatePersistPeriod" : "PT10M",
"maxRowsInMemory" : "100000"
}
},
"properties" : {
"task.partitions" : "1",
"task.replicants" : "1"
}
}
],
"properties" : {
"zookeeper.connect" : "localhost",
"http.port" : "8200",
"http.threads" : "8"
}
}
Then, I start the server using
bin/tranquility server -configFile conf/server.json
I perform post to http://xx.xxx.xxx.xxx:8200/v1/post/wikipedia, with content-type equals application/json
{"timestamp": "2013-08-10T01:02:33Z", "page": "Selamat Pagi", "language" : "en", "user" : "catty", "unpatrolled" : "true", "newPage" : "true", "robot": "false", "anonymous": "false", "namespace":"article", "continent":"North America", "country":"United States", "region":"Bay Area", "city":"San Francisco", "added": 57, "deleted": 200, "delta": -143}
I get the the following respond
{"result":{"received":1,"sent":0}}
It seems that tranquility has received our data, but failed to send it to druid!
I try to run curl -X POST 'http://localhost:8084/druid/v2/?pretty' -H 'content-type: application/json' -d #select.json, but doesn't get the output I inserted via tranquility.
Any idea why? Thanks.
This generally happens when the data you send is out of the window period. If you are inserting data manually, give the exact current timestamp (UTC) in milliseconds. Else it can be easily done if you are using any script to generate data. Make sure it is UTC current time.
It is extremely difficult to setup druid to work properly with real-time data insertion.
The best bet I found is, use https://github.com/implydata . Imply is a set of wrappers around druid, to make it easy to use.
However, the real-time insertion in imply is not perfect either. I had experiment OutOfMemoryException, after inserting 30 millions items via real-time. This will caused data loss on previous inserted 30 millions rows.
The detailed regarding data loss can be found here : https://groups.google.com/forum/#!topic/imply-user-group/95xpYojxiOg
An issue ticket has been filed : https://github.com/implydata/distribution/issues/8
Druid streaming windowPeriod is very short (10 minutes). Outside this period, your event will be ignored.
As you got {"result":{"received":1,"sent":0}}, your worker threads are working fine. Tranquility decides what data is sent to the druid based on the timestamp associated with the data.
This period is decided by the "windowPeriod" configuration. So if your type is realtime ("type":"realtime") and window period is PT10M ("windowPeriod" : "PT10M"), tranquility will send any data between t-10, t+10 and not send anything outside this period.
I disagree with the insertion efficiency problems, we have been sending 3million rows every 15 minutes since June 2016 and has been running beautifully. Of course, we have a stronger infrastructure deemed for the scale.
Another reason for not inserting, is out memory on the coordinador/overloard are running