New Amazon AMIs for EC2 script
Hyy all, I have been using the EC2 script to launch R&D pyspark clusters for a while now. As we use alot of packages such as numpy and scipy with openblas, scikit-learn, bokeh, vowpal wabbit, pystan and etc... All this time, we have been building AMIs on top of the standard spark-AMIs at https://github.com/amplab/spark-ec2/tree/branch-1.6/ami-list/us-east-1 Mainly, I have done the following: - updated yum - Changed the standard python to python 2.7 - changed pip to 2.7 and installed alot of libararies on top of the existing AMIs and created my own AMIs to avoid having to boostrap. But the ec-2 standard AMIs are from *Early February , 2014* and now have become extremely fragile. For example, when I update a certain library, ipython would break, or pip would break and so forth. Can someone please direct me to a more upto date AMI that I can use with more confidence. And I am also interested to know what things need to be in the AMI, if I wanted to build an AMI from scratch (Last resort :( ) And isn't it time to have a ticket in the spark project to build a new suite of AMIs for the EC2 script? https://issues.apache.org/jira/browse/SPARK-922 Many thanks in4maniac -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/New-Amazon-AMIs-for-EC2-script-tp28419.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe e-mail: user-unsubscr...@spark.apache.org
listening to recursive folder structures in s3 using pyspark streaming (textFileStream)
Hi all, I am new to pyspark streaming and I was following a tutorial I saw in the internet (https://github.com/apache/spark/blob/master/examples/src/main/python/streaming/network_wordcount.py). But I replaced the data input with an s3 directory path as: lines = ssc.textFileStream("s3n://bucket/first/second/third1/") When I run the code and upload a file to s3n://bucket/first/second/third1/ (such as s3n://bucket/first/second/third1/test1.txt), the file gets processed as expected. Now I want it to listen to multiple directories and process files if they get uploaded to any of the directories: for example : [s3n://bucket/first/second/third1/, s3n://bucket/first/second/third2/ and s3n://bucket/first/second/third3/] I tried to use the pattern similar to sc.TextFile as : lines = ssc.textFileStream("s3n://bucket/first/second/*/") But this actually didn't work. Can someone please explain to me how I could achieve my objective? thanks in advance !!! in4maniac -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/listening-to-recursive-folder-structures-in-s3-using-pyspark-streaming-textFileStream-tp26247.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: Loading file content based on offsets into the memory
As far as I know, that is not possible. If the file is too big to load to one node, What I would do is to use a RDD.map() function instead to load the file to distributed memory and then filter the lines that are relevant to me. I am not sure how to just read part of a single file. Sorry I'm unable to help here :( -in4 -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Loading-file-content-based-on-offsets-into-the-memory-tp22802p22836.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: AWS-Credentials fails with org.apache.hadoop.fs.s3.S3Exception: FORBIDDEN
HI GUYS... I realised that it was a bug in my code that caused the code to break.. I was running the filter on a SchemaRDD when I was supposed to be running it on an RDD. But I still don't understand why the stderr was about S3 request rather than a type checking error such as "No tuple position 0 found in Row type" was thrown. The error was kinda misleading that I kindof oversaw this logical error in my code. Just thought should keep this posted. -in4 -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/AWS-Credentials-fails-with-org-apache-hadoop-fs-s3-S3Exception-FORBIDDEN-tp22800p22815.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: Loading file content based on offsets into the memory
When loading multiple files, spark loads each file as a partition(block). You can run a function on each partition by using rdd.mapPartitions(function) function. I think you can write a funciton x that extracts everything after the offset and use this funtion with mapPartitions to extract the relevant lines for each file. Hope this helps -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Loading-file-content-based-on-offsets-into-the-memory-tp22802p22804.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
Re: Spark 1.3.1 and Parquet Partitions
Hi V, I am assuming that each of the three .parquet paths you mentioned have multiple partitions in them. For eg: [/dataset/city=London/data.parquet/part-r-0.parquet, /dataset/city=London/data.parquet/part-r-1.parquet] I haven't personally used this with "hdfs", but I've worked with a similar file strucutre with '=' in "S3". And how i get around this is by building a string of all the filepaths seperated by commas (with NO spaces inbetween). Then I use that string as the filepath parameter. I think the following adaptation of S3 file access pattern to HDFS would work If I want to load 1 file: sqlcontext.parquetFile( "hdfs://some ip:8029/dataset/city=London/data.parquet") If I want to load multiple files (lets say all 3 of them): sqlcontext.parquetFile( "hdfs://some ip:8029/dataset/city=London/data.parquet,hdfs://some ip:8029/dataset/city=NewYork/data.parquet,hdfs://some ip:8029/dataset/city=Paris/data.parquet") *** But in the multiple file scenario, the schema of all the files should be the same I hope you can use this S3 pattern with HDFS and hope it works !! Thanks in4 -- View this message in context: http://apache-spark-user-list.1001560.n3.nabble.com/Spark-1-3-1-and-Parquet-Partitions-tp22792p22801.html Sent from the Apache Spark User List mailing list archive at Nabble.com. - To unsubscribe, e-mail: user-unsubscr...@spark.apache.org For additional commands, e-mail: user-h...@spark.apache.org
AWS-Credentials fails with org.apache.hadoop.fs.s3.S3Exception: FORBIDDEN
Hi Guys, I think this problem is related to : http://apache-spark-user-list.1001560.n3.nabble.com/AWS-Credentials-for-private-S3-reads-td8689.html I am running pyspark 1.2.1 in AWS with my AWS credentials exported to master node as Environmental Variables. Halfway through my application, I get thrown with a org.apache.hadoop.fs.s3.S3Exception: org.jets3t.service.S3ServiceException: S3 HEAD request failed for "file path" - ResponseCode=403, ResponseMessage=Forbidden Here is some important information about my job: + my AWS credentials exported to master node as Environmental Variables + there are no '/'s in my secret key + The earlier steps that uses this parquet file actually complete successsfully + The step before the count() does the following: + reads the parquet file (SELECT STATEMENT) + maps it to an RDD + runs a filter on the RDD + The filter works as follows: + extracts one field from each RDD line + checks with a list of 40,000 hashes for presence (if field in LIST_OF_HASHES.value) + LIST_OF_HASHES is a broadcast object The wierdness is that I am using this parquet file in earlier steps and it works fine. The other confusion I have is due to the fact that it only starts failing halfway through the stage. It completes a fraction of tasks and then starts failing.. Hoping to hear something positive. Many thanks in advance Sahanbull The stack trace is as follows: >>> negativeObs.count() [Stage 9:==> (161 + 240) / 800] 15/05/07 07:55:59 ERROR TaskSetManager: Task 277 in stage 9.0 failed 4 times; aborting job Traceback (most recent call last): File "", line 1, in File "/root/spark/python/pyspark/rdd.py", line 829, in count return self.mapPartitions(lambda i: [sum(1 for _ in i)]).sum() File "/root/spark/python/pyspark/rdd.py", line 820, in sum return self.mapPartitions(lambda x: [sum(x)]).reduce(operator.add) File "/root/spark/python/pyspark/rdd.py", line 725, in reduce vals = self.mapPartitions(func).collect() File "/root/spark/python/pyspark/rdd.py", line 686, in collect bytesInJava = self._jrdd.collect().iterator() File "/root/spark/python/lib/py4j-0.8.2.1-src.zip/py4j/java_gateway.py", line 538, in __call__ File "/root/spark/python/lib/py4j-0.8.2.1-src.zip/py4j/protocol.py", line 300, in get_return_value py4j.protocol.Py4JJavaError: An error occurred while calling o139.collect. : org.apache.spark.SparkException: Job aborted due to stage failure: Task 277 in stage 9.0 failed 4 times, most recent failure: Lost task 277.3 in stage 9.0 (TID 4832, ip-172-31-1-185.ec2.internal): org.apache.hadoop.fs.s3.S3Exception: org.jets3t.service.S3ServiceException: S3 HEAD request failed for '/subbucket%2Fpath%2F2Fpath%2F2Fpath%2F2Fpath%2F2Fpath%2Ffilename.parquet%2Fpart-r-349.parquet' - ResponseCode=403, ResponseMessage=Forbidden at org.apache.hadoop.fs.s3native.Jets3tNativeFileSystemStore.retrieveMetadata(Jets3tNativeFileSystemStore.java:122) at sun.reflect.GeneratedMethodAccessor116.invoke(Unknown Source) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) at java.lang.reflect.Method.invoke(Method.java:606) at org.apache.hadoop.io.retry.RetryInvocationHandler.invokeMethod(RetryInvocationHandler.java:82) at org.apache.hadoop.io.retry.RetryInvocationHandler.invoke(RetryInvocationHandler.java:59) at org.apache.hadoop.fs.s3native.$Proxy9.retrieveMetadata(Unknown Source) at org.apache.hadoop.fs.s3native.NativeS3FileSystem.getFileStatus(NativeS3FileSystem.java:326) at parquet.hadoop.ParquetFileReader.readFooter(ParquetFileReader.java:381) at parquet.hadoop.ParquetRecordReader.initializeInternalReader(ParquetRecordReader.java:155) at parquet.hadoop.ParquetRecordReader.initialize(ParquetRecordReader.java:138) at org.apache.spark.rdd.NewHadoopRDD$$anon$1.(NewHadoopRDD.scala:135) at org.apache.spark.rdd.NewHadoopRDD.compute(NewHadoopRDD.scala:107) at org.apache.spark.rdd.NewHadoopRDD.compute(NewHadoopRDD.scala:69) at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:280) at org.apache.spark.rdd.RDD.iterator(RDD.scala:247) at org.apache.spark.rdd.MappedRDD.compute(MappedRDD.scala:31) at org.apache.spark.sql.SchemaRDD.compute(SchemaRDD.scala:120) at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:280) at org.apache.spark.rdd.RDD.iterator(RDD.scala:247) at org.apache.spark.rdd.MappedRDD.compute(MappedRDD.scala:31) at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:280) at org.apache.spark.rdd.RDD.iterator(RDD.scala:247) at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:35) at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:280) at org.apache.spark.rdd.RDD.iterator(RDD.scala:247) at org.apache.spark.api.python.