I want to apply custom logic over dataset placed in Redshift.
Example of input data:
userid, event, fileid, timestamp, ....
100000, start, 120, 2018-09-17 19:11:40
100000, done, 120, 2018-09-17 19:12:40
100000, done, 120, 2018-09-17 19:13:40
100000, start, 500, 2018-09-17 19:13:50
100000, done, 120, 2018-09-17 19:14:40
100000, done, 500, 2018-09-17 19:14:50
100000, done, 120, 2018-09-17 19:15:40
This means something like:
file 120: start-----done-----done-----done-----done
file 150: start-----done
time : 11:40----12:40----13:40-----14:40-----15:40
But it should looks like
file 120: start-----done-----done
file 150: start-----done
time : 11:40----12:40----13:40-----14:40-----15:40
The file 120 has been interrupted once the file 150 has been started
Keep in mind that a lot if different users here and many different files.
Cleaned data should be:
userid, event, fileid, timestamp, ....
100000, start, 120, 2018-09-17 19:11:40
100000, done, 120, 2018-09-17 19:12:40
100000, done, 120, 2018-09-17 19:13:40
100000, start, 500, 2018-09-17 19:13:50
100000, done, 500, 2018-09-17 19:14:50
It shouldn't be able to have multiple concurrent files at once for same user. So after second one has started, events from first one should not be removed from current dataset.
The code is simple but on python and it's easy scalable for Google Dataflow, for example, but moving 100GB+ from AWS to GC is not good idea.
Question #1:
Is it possible to do it on SQL (using postgres/redshift specific features) or better to use Spark? (but not sure how to implement it there)
Question #2:
Any suggestion on maybe better to use AWS Batch or whatever, cause with apache beam - it's easy and pretty much obvious, but how AWS Batch works and how to divide the dataset on chunks (like group per user) - it's a big question.
My suggestion is to somehow unload data from redshift into S3 bucket but divide it in manner separate file=user, then if aws batch supporting this - just feed the bucket and each file should be processed concurrently on already created instances. Not sure if this is makes sense.
If you want to remove rows where the fileid does not match the most recent start for the user, you can use lag(ignore nulls):
select t.*
from (select t.*,
lag(case when event = 'start' then file_id end ignore nulls) over (partition by userid order by timestamp) as start_fileid
from t
) t
where event = 'start' or start_fileid = fileid;
Related
I want to continue the DAG only if a csv file exists in S3, otherwise it should just end.
The DAG itself is being scheduled hourly.
with DAG(dag_id="my_dag",
start_date=datetime(2023, 1, 1),
schedule_interval='#hourly',
catchup=False
) as dag:
check_for_new_csv = S3KeySensor(
task_id='check_for_new_csv',
bucket_name='bucket-data',
bucket_key='*.csv',
wildcard_match=True,
soft_fail=True,
retries=1
)
start_instance = EC2StartInstanceOperator(
task_id="start_ec2_instance_task",
instance_id=INSTANCE_ID,
region_name=REGION
)
check_for_new_csv >> start_instance
But the sensor seems to run forever - in the log I can see it keeps on running:
[2023-01-10, 15:02:06 UTC] {s3.py:98} INFO - Poking for key : s3://bucket-data/*.csv
[2023-01-10, 15:03:08 UTC] {s3.py:98} INFO - Poking for key : s3://bucket-data/*.csv
Maybe the sensor in not the best choice for such logic?
A sensor is a perfect choice for this use case. I'd try setting the poke_interval and timeout to different smaller values than their default to make sure the sensor that Airflow is checking on the right intervals (by default, they are very long).
One thing to watch out for is if your sensors run on longer intervals than your schedule interval. For example, if your DAG is scheduled to run hourly, but your sensors's timeout is set for 2 hours, your next DAG run may not run as expected (depending on your concurrency and max_active_dag settings), or it may run unexpectedly because the sensor detects an older file. Ideally, you can append a timestamp in the name of your file to avoid this.
I have several projects created on the web interface, each has several batches that are already ended. The max assignment per task is 3. I need to add 2 more assignments for each HIT, is it possible?
I've tried using the API on a in-progress batch:
mturk.create_additional_assignments_for_hit(HITId=HIT_ID,NumberOfAdditionalAssignments=2)
and the response :
{'ResponseMetadata': {'RequestId': '.....some id ...',
'HTTPStatusCode': 200,
'HTTPHeaders': {'x-amzn-requestid': '.....some id ...',
'content-type': 'application/x-amz-json-1.1',
'content-length': '2',
'date': 'Thu, 20 Jan 2022 12:20:02 GMT'},
'RetryAttempts': 0}}
But I can't see any update on the web for +2 extra assignments..
Found the answer.
First, the Requester UI (RUI) and the API are not fully connected, such that not all changes in the API will be visible in the RUI.
Here is my answer using the python API
To "revive" old HITs and add a new assignments to them you need to:
mturk = boto3.client('mturk',
aws_access_key_id='xxxxxxxxxxxx',
aws_secret_access_key='xxxxxxxxxxxxx',
region_name='us-east-1',
endpoint_url='https://mturk-requester.us-east-1.amazonaws.com')
Extend the expiration date for the HIT (even if its already passed):
mturk.update_expiration_for_hit(HITId=HIT_ID_STRING,ExpireAt=datetime.datetime(2022, 1, 23, 20, 00, 00))
Then, increase max assignments, here we add +2 more:
mturk.create_additional_assignments_for_hit(HITId=HIT_ID_STRING,NumberOfAdditionalAssignments=2)
That's it, you can see that the total number of NumberOfAssignmentsAvailable increased by 2 and that MaxAssignments increased as well:
mturk.get_hit(HITId=HIT_ID_STRING)
'MaxAssignments':5,
'NumberOfAssignmentsPending': 0,
'NumberOfAssignmentsAvailable': 2,
'NumberOfAssignmentsCompleted': 3
I have a few jobs, say one is loading a text file from a google cloud storage bucket to bigquery table, and another one is a scheduled query to copy data from one table to another table with some transformation, I want the second job to depend on the success of the first one, how do we achieve this in bigquery if it is possible to do so at all?
Many thanks.
Best regards,
Right now a developer needs to put together the chain of operations.
It can be done either using Cloud Functions (supports, Node.js, Go, Python) or via Cloud Run container (supports gcloud API, any programming language).
Basically you need to
issue a job
get the job id
poll for the job id
job is finished trigger other steps
If using Cloud Functions
place the file into a dedicated GCS bucket
setup a GCF that monitors that bucket and when a new file is uploaded it will execute a function that imports into GCS - wait until the operations ends
at the end of the GCF you can trigger other functions for next step
another use case with Cloud Functions:
A: a trigger starts the GCF
B: function executes the query (copy data to another table)
C: gets a job id - fires another function with a bit of delay
I: a function gets a jobid
J: polls for job is ready?
K: if not ready, fires himself again with a bit of delay
L: if ready triggers next step - could be a dedicated function or parameterized function
It is possible to address your scenario with either cloud functions(CF) or with a scheduler (airflow). The first approach is event-driven getting your data crunch immediately. With the scheduler, expect data availability delay.
As it has been stated once you submit BigQuery job you get back job ID, that needs to be check till it completes. Then based on the status you can handle on success or failure post actions respectively.
If you were to develop CF, note that there are certain limitations like execution time (max 9min), which you would have to address in case BigQuery job takes more than 9 min to complete. Another challenge with CF is idempotency, making sure that if the same datafile event comes more than once, the processing should not result in data duplicates.
Alternatively, you can consider using some event-driven serverless open source projects like BqTail - Google Cloud Storage BigQuery Loader with post-load transformation.
Here is an example of the bqtail rule.
rule.yaml
When:
Prefix: "/mypath/mysubpath"
Suffix: ".json"
Async: true
Batch:
Window:
DurationInSec: 85
Dest:
Table: bqtail.transactions
Transient:
Dataset: temp
Alias: t
Transform:
charge: (CASE WHEN type_id = 1 THEN t.payment + f.value WHEN type_id = 2 THEN t.payment * (1 + f.value) END)
SideInputs:
- Table: bqtail.fees
Alias: f
'On': t.fee_id = f.id
OnSuccess:
- Action: query
Request:
SQL: SELECT
DATE(timestamp) AS date,
sku_id,
supply_entity_id,
MAX($EventID) AS batch_id,
SUM( payment) payment,
SUM((CASE WHEN type_id = 1 THEN t.payment + f.value WHEN type_id = 2 THEN t.payment * (1 + f.value) END)) charge,
SUM(COALESCE(qty, 1.0)) AS qty
FROM $TempTable t
LEFT JOIN bqtail.fees f ON f.id = t.fee_id
GROUP BY 1, 2, 3
Dest: bqtail.supply_performance
Append: true
OnFailure:
- Action: notify
Request:
Channels:
- "#e2e"
Title: Failed to aggregate data to supply_performance
Message: "$Error"
OnSuccess:
- Action: query
Request:
SQL: SELECT CURRENT_TIMESTAMP() AS timestamp, $EventID AS job_id
Dest: bqtail.supply_performance_batches
Append: true
- Action: delete
You want to use an orchestration tool, especially if you want to set up this tasks as recurring jobs.
We use Google Cloud Composer, which is a managed service based on Airflow, to do workflow orchestration and works great. It comes with automatically retry, monitoring, alerting, and much more.
You might want to give it a try.
Basically you can use Cloud Logging to know almost all kinds of operations in GCP.
BigQuery is no exception. When the query job completed, you can find the corresponding log in the log viewer.
The next question is how to anchor the exact query you want, one way to achieve this is to use labeled query (means attach labels to your query) [1].
For example, you can use below bq command to issue query with foo:bar label
bq query \
--nouse_legacy_sql \
--label foo:bar \
'SELECT COUNT(*) FROM `bigquery-public-data`.samples.shakespeare'
Then, when you go to Logs Viewer and issue below log filter, you will find the exactly log generated by above query.
resource.type="bigquery_resource"
protoPayload.serviceData.jobCompletedEvent.job.jobConfiguration.labels.foo="bar"
The next question is how to emit an event based on this log for the next workload. Then, the Cloud Pub/Sub comes into play.
2 ways to publish an event based on log pattern are:
Log Routers: set Pub/Sub topic as the destination [1]
Log-based Metrics: create alert policy whose notification channel is Pub/Sub [2]
So, the next workload can subscribe to the Pub/Sub topic, and be triggered when the previous query has completed.
Hope this helps ~
[1] https://cloud.google.com/bigquery/docs/reference/rest/v2/Job#jobconfiguration
[2] https://cloud.google.com/logging/docs/routing/overview
[3] https://cloud.google.com/logging/docs/logs-based-metrics
I'm trying to get a streaming aggregation/groupBy working in append output mode, to be able to use the resulting stream in a stream-to-stream join. I'm working on (Py)Spark 2.3.2, and I'm consuming from Kafka topics.
My pseudo-code is something like below, running in a Zeppelin notebook
orderStream = spark.readStream().format("kafka").option("startingOffsets", "earliest").....
orderGroupDF = (orderStream
.withWatermark("LAST_MOD", "20 seconds")
.groupBy("ID", window("LAST_MOD", "10 seconds", "5 seconds"))
.agg(
collect_list(struct("attra", "attrb2",...)).alias("orders"),
count("ID").alias("number_of_orders"),
sum("PLACED").alias("number_of_placed_orders"),
min("LAST_MOD").alias("first_order_tsd")
)
)
debug = (orderGroupDF.writeStream
.outputMode("append")
.format("memory").queryName("debug").start()
)
After that, I would expected that data appears on the debug query and I can select from it (after the late arrival window of 20 seconds has expired. But no data every appears on the debug query (I waited several minutes)
When I changed output mode to update the query works immediately.
Any hint what I'm doing wrong?
EDIT: after some more experimentation, I can add the following (but I still don't understand it).
When starting the Spark application, there is quite a lot of old data (with event timestamps << current time) on the topic from which I consume. After starting, it seems to read all these messages (MicroBatchExecution in the log reports "numRowsTotal = 6224" for example), but nothing is produced on the output, and the eventTime watermark in the log from MicroBatchExecution stays at epoch (1970-01-01).
After producing a fresh message onto the input topic with eventTimestamp very close to current time, the query immediately outputs all the "queued" records at once, and bumps the eventTime watermark in the query.
What I can also see that there seems to be an issue with the timezone. My Spark programs runs in CET (UTC+2 currently). The timestamps in the incoming Kafka messages are in UTC, e.g "LAST__MOD": "2019-05-14 12:39:39.955595000". I have set spark_sess.conf.set("spark.sql.session.timeZone", "UTC"). Still, the microbatch report after that "new" message has been produced onto the input topic says
"eventTime" : {
"avg" : "2019-05-14T10:39:39.955Z",
"max" : "2019-05-14T10:39:39.955Z",
"min" : "2019-05-14T10:39:39.955Z",
"watermark" : "2019-05-14T10:35:25.255Z"
},
So the eventTime somehow links of with the time in the input message, but it is 2 hours off. The UTC difference has been subtraced twice. Additionally, I fail to see how the watermark calculation works. Given that I set it to 20 seconds, I would have expected it to be 20 seconds older than the max eventtime. But apparently it is 4 mins 14 secs older. I fail to see the logic behind this.
I'm very confused...
It seems that this was related to the Spark version 2.3.2 that I used, and maybe more concretely to SPARK-24156. I have upgraded to Spark 2.4.3 and here I get the results of the groupBy immediately (well, of course after the watermark lateThreshold has expired, but "in the expected timeframe".
I am running Spark in standalone mode on 2 machines which have these configs
500gb memory, 4 cores, 7.5 RAM
250gb memory, 8 cores, 15 RAM
I have created a master and a slave on 8core machine, giving 7 cores to worker. I have created another slave on 4core machine with 3 worker cores. The UI shows 13.7 and 6.5 G usable RAM for 8core and 4core respectively.
Now on this I have to process an aggregate of user ratings over a period of 15 days. I am trying to do this using Pyspark
This data is stored in hourwise files in day-wise directories in an s3 bucket, every file must be around 100MB eg
s3://some_bucket/2015-04/2015-04-09/data_files_hour1
I am reading the files like this
a = sc.textFile(files, 15).coalesce(7*sc.defaultParallelism) #to restrict partitions
where files is a string of this form 's3://some_bucket/2015-04/2015-04-09/*,s3://some_bucket/2015-04/2015-04-09/*'
Then I do a series of maps and filters and persist the result
a.persist(StorageLevel.MEMORY_ONLY_SER)
Then I need to do a reduceByKey to get an aggregate score over the span of days.
b = a.reduceByKey(lambda x, y: x+y).map(aggregate)
b.persist(StorageLevel.MEMORY_ONLY_SER)
Then I need to make a redis call for the actual terms for the items the user has rated, so I call mapPartitions like this
final_scores = b.mapPartitions(get_tags)
get_tags function creates a redis connection each time of invocation and calls redis and yield a (user, item, rate) tuple
(The redis hash is stored in the 4core)
I have tweaked the settings for SparkConf to be at
conf = (SparkConf().setAppName(APP_NAME).setMaster(master)
.set("spark.executor.memory", "5g")
.set("spark.akka.timeout", "10000")
.set("spark.akka.frameSize", "1000")
.set("spark.task.cpus", "5")
.set("spark.cores.max", "10")
.set("spark.serializer", "org.apache.spark.serializer.KryoSerializer")
.set("spark.kryoserializer.buffer.max.mb", "10")
.set("spark.shuffle.consolidateFiles", "True")
.set("spark.files.fetchTimeout", "500")
.set("spark.task.maxFailures", "5"))
I run the job with driver-memory of 2g in client mode, since cluster mode doesn't seem to be supported here.
The above process takes a long time for 2 days' of data (around 2.5hours) and completely gives up on 14 days'.
What needs to improve here?
Is this infrastructure insufficient in terms of RAM and cores (This is offline and can take hours, but it has got to finish in 5 hours or so)
Should I increase/decrease the number of partitions?
Redis could be slowing the system, but the number of keys is just too huge to make a one time call.
I am not sure where the task is failing, in reading the files or in reducing.
Should I not use Python given better Spark APIs in Scala, will that help with efficiency as well?
This is the exception trace
Lost task 4.1 in stage 0.0 (TID 11, <node>): java.net.SocketTimeoutException: Read timed out
at java.net.SocketInputStream.socketRead0(Native Method)
at java.net.SocketInputStream.read(SocketInputStream.java:152)
at java.net.SocketInputStream.read(SocketInputStream.java:122)
at sun.security.ssl.InputRecord.readFully(InputRecord.java:442)
at sun.security.ssl.InputRecord.readV3Record(InputRecord.java:554)
at sun.security.ssl.InputRecord.read(InputRecord.java:509)
at sun.security.ssl.SSLSocketImpl.readRecord(SSLSocketImpl.java:934)
at sun.security.ssl.SSLSocketImpl.readDataRecord(SSLSocketImpl.java:891)
at sun.security.ssl.AppInputStream.read(AppInputStream.java:102)
at org.apache.http.impl.io.AbstractSessionInputBuffer.read(AbstractSessionInputBuffer.java:198)
at org.apache.http.impl.io.ContentLengthInputStream.read(ContentLengthInputStream.java:178)
at org.apache.http.impl.io.ContentLengthInputStream.read(ContentLengthInputStream.java:200)
at org.apache.http.impl.io.ContentLengthInputStream.close(ContentLengthInputStream.java:103)
at org.apache.http.conn.BasicManagedEntity.streamClosed(BasicManagedEntity.java:164)
at org.apache.http.conn.EofSensorInputStream.checkClose(EofSensorInputStream.java:227)
at org.apache.http.conn.EofSensorInputStream.close(EofSensorInputStream.java:174)
at org.apache.http.util.EntityUtils.consume(EntityUtils.java:88)
at org.jets3t.service.impl.rest.httpclient.HttpMethodReleaseInputStream.releaseConnection(HttpMethodReleaseInputStream.java:102)
at org.jets3t.service.impl.rest.httpclient.HttpMethodReleaseInputStream.close(HttpMethodReleaseInputStream.java:194)
at org.apache.hadoop.fs.s3native.NativeS3FileSystem$NativeS3FsInputStream.seek(NativeS3FileSystem.java:152)
at org.apache.hadoop.fs.BufferedFSInputStream.seek(BufferedFSInputStream.java:89)
at org.apache.hadoop.fs.FSDataInputStream.seek(FSDataInputStream.java:63)
at org.apache.hadoop.mapred.LineRecordReader.<init>(LineRecordReader.java:126)
at org.apache.hadoop.mapred.TextInputFormat.getRecordReader(TextInputFormat.java:67)
at org.apache.spark.rdd.HadoopRDD$$anon$1.<init>(HadoopRDD.scala:236)
at org.apache.spark.rdd.HadoopRDD.compute(HadoopRDD.scala:212)
at org.apache.spark.rdd.HadoopRDD.compute(HadoopRDD.scala:101)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:277)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:244)
at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:35)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:277)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:244)
at org.apache.spark.rdd.CoalescedRDD$$anonfun$compute$1.apply(CoalescedRDD.scala:93)
at org.apache.spark.rdd.CoalescedRDD$$anonfun$compute$1.apply(CoalescedRDD.scala:92)
at scala.collection.Iterator$$anon$13.hasNext(Iterator.scala:371)
at scala.collection.Iterator$class.foreach(Iterator.scala:727)
at scala.collection.AbstractIterator.foreach(Iterator.scala:1157)
at org.apache.spark.api.python.PythonRDD$.writeIteratorToStream(PythonRDD.scala:405)
at org.apache.spark.api.python.PythonRDD$WriterThread$$anonfun$run$1.apply(PythonRDD.scala:243)
at org.apache.spark.util.Utils$.logUncaughtExceptions(Utils.scala:1617)
at org.apache.spark.api.python.PythonRDD$WriterThread.run(PythonRDD.scala:205)
I could really use some help, thanks in advance
Here is what my main code looks like
def main(sc):
f=get_files()
a=sc.textFile(f, 15)
.coalesce(7*sc.defaultParallelism)
.map(lambda line: line.split(","))
.filter(len(line)>0)
.map(lambda line: (line[18], line[2], line[13], line[15])).map(scoring)
.map(lambda line: ((line[0], line[1]), line[2])).persist(StorageLevel.MEMORY_ONLY_SER)
b=a.reduceByKey(lambda x, y: x+y).map(aggregate)
b.persist(StorageLevel.MEMORY_ONLY_SER)
c=taggings.mapPartitions(get_tags)
c.saveAsTextFile("f")
a.unpersist()
b.unpersist()
The get_tags function is
def get_tags(partition):
rh = redis.Redis(host=settings['REDIS_HOST'], port=settings['REDIS_PORT'], db=0)
for element in partition:
user = element[0]
song = element[1]
rating = element[2]
tags = rh.hget(settings['REDIS_HASH'], song)
if tags:
tags = json.loads(tags)
else:
tags = scrape(song, rh)
if tags:
for tag in tags:
yield (user, tag, rating)
The get_files function is as:
def get_files():
paths = get_path_from_dates(DAYS)
base_path = 's3n://acc_key:sec_key#bucket/'
files = list()
for path in paths:
fle = base_path+path+'/file_format.*'
files.append(fle)
return ','.join(files)
The get_path_from_dates(DAYS) is
def get_path_from_dates(last):
days = list()
t = 0
while t <= last:
d = today - timedelta(days=t)
path = d.strftime('%Y-%m')+'/'+d.strftime('%Y-%m-%d')
days.append(path)
t += 1
return days
As a small optimization, I have created two separate tasks, one to read from s3 and get additive sum, second to read transformations from redis. The first tasks has high number of partitions since there are around 2300 files to read. The second one has much lesser number of partitions to prevent redis connection latency, and there is only one file to read which is on the EC2 cluster itself. This is only partial, still looking for suggestions to improve ...
I was in a similar usecase: doing coalesce on a RDD with 300,000+ partitions. The difference is that I was using s3a(SocketTimeoutException from S3AFileSystem.waitAysncCopy). Finally the issue was resolved by setting a larger fs.s3a.connection.timeout(Hadoop's core-site.xml). Hopefully you can get a clue.