COS418 Assignment 1: Sequential Map/Reduce

Introduction

In the first two assignments, you will build a Map/Reduce library as a way to learn the Go programming language and as a way to learn about fault tolerance in distributed systems. For Assignment 1, you will work with a sequential Map/Reduce implementation and write a sample program that uses it. The interface to the library is similar to the one described in the original MapReduce paper.

Software

You'll implement this assignment (and all the assignments) in Go. The Go web site contains lots of tutorial information which you may want to look at.

For each assignment, we will provide you with a significant amount of scaffolding code to get started. You can download a tarball containing the full repository of provided code for all assignments here. We will ensure that all the code we supply works on the CS servers (cycles.cs.princeton.edu). We expect that it is likely to work on your own development environment that supports Go. You can use the instructions provided to configure your own environment.

For the first two assignments, we supply you with parts of a flexible MapReduce implementation. It has support for two modes of operation, sequential and distributed. Assignment 1 deals with the former. The map and reduce tasks are all executed in serial: the first map task is executed to completion, then the second, then the third, etc. When all the map tasks have finished, the first reduce task is run, then the second, etc. This mode, while not very fast, can be very useful for debugging, since it removes much of the noise seen in a parallel execution. The sequential mode also simplifies or eliminates various corner cases of a distributed system.

Getting familiar with the source

The mapreduce package (located at $GOPATH/src/mapreduce) provides a simple Map/Reduce library with a sequential implementation. Applications would normally call Distributed() — located in mapreduce/master.go — to start a job, but may instead call Sequential() — also in mapreduce/master.go — to get a sequential execution, which will be our approach in this assignment.

The flow of the mapreduce implementation is as follows:

  1. The application provides a number of input files, a map function, a reduce function, and the number of reduce tasks (nReduce).
  2. A master is created with this knowledge. It spins up an RPC server (see mapreduce/master_rpc.go), and waits for workers to register (using the RPC call Register() defined in mapreduce/master.go). As tasks become available, schedule() — located in mapreduce/schedule.go — decides how to assign those tasks to workers, and how to handle worker failures.
  3. The master considers each input file one map task, and makes a call to doMap() in mapreduce/common_map.go at least once for each task. It does so either directly (when using Sequential()) or by issuing the DoTask RPC — located in mapreduce/worker.go — on a worker. Each call to doMap() reads the appropriate file, calls the map function on that file's contents, and produces nReduce files for each map file. Thus, after all map tasks are done, the total number of files will be the product of the number of files given to map (nIn) and nReduce.
    f0-0, ...,  f0-[nReduce-1],
    ...
    f[nIn-1]-0, ..., f[nIn-1]-[nReduce-1].
  4. The master next makes a call to doReduce() in mapreduce/common_reduce.go at least once for each reduce task. As with doMap(), it does so either directly or through a worker. doReduce() collects corresponding files from each map result (e.g. f0-i, f1-i, ... f[nIn-1]-i), and runs the reduce function on each collection. This process produces nReduce result files.
  5. The master calls mr.merge() in mapreduce/master_splitmerge.go, which merges all the nReduce files produced by the previous step into a single output.
  6. The master sends a Shutdown RPC to each of its workers, and then shuts down its own RPC server.
You should look through the files in the MapReduce implementation, as reading them might be useful to understand how the other methods fit into the overall architecture of the system hierarchy. However, for this assignment, you will write/modify only doMap in mapreduce/common_map.go, doReduce in mapreduce/common_reduce.go, and mapF and reduceF in main/wc.go. You will not be able to submit other files or modules.

Part I: Map/Reduce input and output

The Map/Reduce implementation you are given is missing some pieces. Before you can write your first Map/Reduce function pair, you will need to fix the sequential implementation. In particular, the code we give you is missing two crucial pieces: the function that divides up the output of a map task, and the function that gathers all the inputs for a reduce task. These tasks are carried out by the doMap() function in mapreduce/common_map.go, and the doReduce() function in mapreduce/common_reduce.go respectively. The comments in those files should point you in the right direction.

To help you determine if you have correctly implemented doMap() and doReduce(), we have provided you with a Go test suite that checks the correctness of your implementation. These tests are implemented in the file test_test.go. To run the tests for the sequential implementation that you have now fixed, follow this (or a non-bash equivalent) sequence of shell commands, starting from the cos418 top-level directory (that is, the top level of the hierarchy contained in the tarball you downloaded above):

$ cd 418 # or whatever you have called the directory that you unpacked from the tarball
$ export GOPATH="$PWD"  # go needs $GOPATH to be set to the project's directory, above src.
$ cd "$GOPATH/src/mapreduce"
$ go test -run Sequential mapreduce/...
ok  mapreduce 4.515s

If the output did not show ok next to the tests, your implementation has a bug in it. To give more verbose output, set debugEnabled = true in mapreduce/common.go, and add -v to the test command above. You will get much more output along the lines of:

$ go test -v -run Sequential
=== RUN   TestSequentialSingle
master: Starting Map/Reduce task test
Merge: read mrtmp.test-res-0
master: Map/Reduce task completed
--- PASS: TestSequentialSingle (2.30s)
=== RUN   TestSequentialMany
master: Starting Map/Reduce task test
Merge: read mrtmp.test-res-0
Merge: read mrtmp.test-res-1
Merge: read mrtmp.test-res-2
master: Map/Reduce task completed
--- PASS: TestSequentialMany (2.32s)
PASS
ok  mapreduce4.635s

Part II: Single-worker word count

Now that the map and reduce tasks are connected, we can start implementing some interesting Map/Reduce operations. For this assignment, we will be implementing word count — a simple and classic Map/Reduce example. Specifically, your task is to modify mapF and reduceF within main/wc.go so that the application reports the number of occurrences of each word. A word is any contiguous sequence of letters, as determined by unicode.IsLetter.

There are some input files with pathnames of the form pg-*.txt in the main directory, downloaded from Project Gutenberg. This is the result when you initially try to compile the code we provide you and run it:

$ cd "$GOPATH/src/main"
$ go run wc.go master sequential pg-*.txt
# command-line-arguments
./wc.go:14: missing return at end of function
./wc.go:21: missing return at end of function

The compilation fails because we haven't written a complete map function (mapF()) nor a complete reduce function (reduceF()) in wc.go yet. Before you start coding read Section 2 of the MapReduce paper. Your mapF() and reduceF() functions will differ a bit from those in the paper's Section 2.1. Your mapF() will be passed the name of a file, as well as that file's contents; it should split it into words, and return a Go slice of key/value pairs, of type mapreduce.KeyValue. Your reduceF() will be called once for each key, with a slice of all the values generated by mapF() for that key; it should return a single output value.

You can test your solution using:

$ cd "$GOPATH/src/main"
$ go run wc.go master sequential pg-*.txt
master: Starting Map/Reduce task wcseq
Merge: read mrtmp.wcseq-res-0
Merge: read mrtmp.wcseq-res-1
Merge: read mrtmp.wcseq-res-2
master: Map/Reduce task completed
The output will be in the file mrtmp.wcseq. We will test your implementation's correctness with the following command, which should produce the following top 10 words:
$ sort -n -k2 mrtmp.wcseq | tail -10
he: 34077
was: 37044
that: 37495
I: 44502
in: 46092
a: 60558
to: 74357
of: 79727
and: 93990
the: 154024
(this sample result is also found in main/mr-testout.txt)

You can remove the output file and all intermediate files with:

$ rm mrtmp.*

To make testing easy for you, from the $GOPATH/src/main directory, run:

$ sh ./test-wc.sh
and it will report if your solution is correct or not.

Resources and Advice

Submission

Submit your code to the COS418 Assignment 1 Dropbox. You may submit multiple times, only the one in the Dropbox at the time of grading will be recorded.

The Dropbox script will run only a small number of tests (though you should of course pass these!). It is not a substitute for doing your own testing on the full go test cases detailed above. Before submitting, please run the full tests given above for both parts one final time. You are responsible for making sure your code works.

You will receive full credit for Part I if your software passes the Sequential tests (as run by the go test commands above) on the CS servers. You will receive full credit for Part II if your Map/Reduce word count output matches the correct output for the sequential execution above when run on the CS servers.

The final portion of your credit is determined by code quality tests, using the standard tools gofmt and go vet. You will receive full credit for this portion if all files submitted conform to the style standards set by gofmt and the report from go vet is clean for your mapreduce package (that is, produces no errors). If your code does not pass the gofmt test, you should reformat your code using the tool. You can also use the Go Checkstyle tool for advice to improve your code's style, if applicable. Additionally, though not part of the graded cheks, it would also be advisable to produce code that complies with Golint where possible.

Acknowledgements

This assignment is adapted from MIT's 6.824 course. Thanks to Frans Kaashoek, Robert Morris, and Nickolai Zeldovich for their support.


Last updated: 2016-09-20 19:55:54 -0400