job_pool 0.5 → 0.6
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.gitignore +3 -1
- data/README.md +18 -7
- data/Rakefile +8 -0
- data/lib/job_pool.rb +3 -0
- data/lib/job_pool/job.rb +59 -31
- data/lib/job_pool/version.rb +1 -1
- data/spec/job_pool/job_spec.rb +32 -33
- data/spec/job_pool_spec.rb +9 -8
- data/spec/readme_spec.rb +6 -4
- metadata +3 -2
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 715c70888317b1db49c5644d30e1197d6cd3dcb6
|
4
|
+
data.tar.gz: 15094b1d5774c55b9bb21d469405c65f7ea28942
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 239d6296f2933a8e63a0747b99f1967720b1bb4137c2322b2bf3b90f0c3e07f521a4822e5ebeddd8e942a66d6bcc7f2e2bd00d62f62a68d3a4f9de588b426109
|
7
|
+
data.tar.gz: 04597306e45e9b2d83538bb1a1d4585ecb8a00f48229a6aeb5df2db7082c6de3645581441d2931941593d2f98633cf0d15b180120ee679aaae7b5d0511b745f9
|
data/.gitignore
CHANGED
data/README.md
CHANGED
@@ -35,7 +35,7 @@ pool = JobPool.new
|
|
35
35
|
Then fire off a job. This one waits a bit and then ROT-13s its input.
|
36
36
|
|
37
37
|
```ruby
|
38
|
-
job = pool.launch("sleep 5; tr A-Za-z N-ZA-Mn-za-m", "the secrets")
|
38
|
+
job = pool.launch("sleep 5; tr A-Za-z N-ZA-Mn-za-m", stdin: "the secrets")
|
39
39
|
pool.count => 1
|
40
40
|
job.output => ""
|
41
41
|
(after five seconds)
|
@@ -47,14 +47,16 @@ job.output => "gur frpergf"
|
|
47
47
|
|
48
48
|
You can specify IO objects to read from and write to:
|
49
49
|
|
50
|
-
TODO: this works, but it closes your stdout! That's problematic.
|
51
|
-
Maybe add a mode that doesn't close the output stream when done?
|
52
|
-
Or just use a different example?
|
53
|
-
|
54
50
|
```ruby
|
55
|
-
|
51
|
+
source = File.open('contents.txt.gz')
|
52
|
+
desdtination = File.open('/tmp/out', 'w')
|
53
|
+
pool.launch 'gunzip --to-stdout', stdin: source, stdout: destination
|
56
54
|
```
|
57
55
|
|
56
|
+
Note that if you specify STDIN or STDOUT, job_pool will close the stream
|
57
|
+
when the child terminates. This is almost certainly not what you want.
|
58
|
+
TODO: add an option... `keep_open: :stdout`, `keep_open: [:stdin, :stdout]`
|
59
|
+
|
58
60
|
#### Killing a Job
|
59
61
|
|
60
62
|
If you want to terminate a job, just kill it:
|
@@ -73,7 +75,7 @@ Pass the number of seconds to wait, default is 2 seconds.
|
|
73
75
|
|
74
76
|
#### Timeouts
|
75
77
|
|
76
|
-
TODO
|
78
|
+
TODO: add a timeout example
|
77
79
|
|
78
80
|
#### Limiting Running Processes
|
79
81
|
|
@@ -103,6 +105,15 @@ TODO: friggin documentation!
|
|
103
105
|
TODO: include an example of a job queue
|
104
106
|
|
105
107
|
|
108
|
+
## Documentation
|
109
|
+
|
110
|
+
I tried to use Ruby's automated documentation tools but it didn't stick.
|
111
|
+
RDoc didn't have reliable markdown support and lacked param name checking.
|
112
|
+
YARD produces uglier output (especially noframes) and its lack of :nodoc:
|
113
|
+
makes my docs too noisy. For now I'll just write documentation how I want
|
114
|
+
to see it and hope some tool catches up.
|
115
|
+
|
116
|
+
|
106
117
|
## License
|
107
118
|
|
108
119
|
MIT, enjoy!
|
data/Rakefile
CHANGED
@@ -1,6 +1,14 @@
|
|
1
1
|
require 'bundler/gem_tasks'
|
2
2
|
require 'rspec/core/rake_task'
|
3
|
+
require 'yard'
|
3
4
|
|
4
5
|
RSpec::Core::RakeTask.new(:spec)
|
5
6
|
task :default => ['spec']
|
6
7
|
task :test => ['spec']
|
8
|
+
|
9
|
+
YARD::Rake::YardocTask.new do |t|
|
10
|
+
t.files = ['lib/**/*.rb']
|
11
|
+
# t.options = ['--any', '--extra', '--opts']
|
12
|
+
t.stats_options = ['--list-undoc']
|
13
|
+
end
|
14
|
+
task :doc => ['yard']
|
data/lib/job_pool.rb
CHANGED
@@ -6,10 +6,13 @@ require 'job_pool/job'
|
|
6
6
|
# TODO: rewrite wait_next
|
7
7
|
|
8
8
|
class JobPool
|
9
|
+
# Indicates that the maximum allowed jobs are already running. Set in [initialize], thrown by [launch].
|
9
10
|
class TooManyJobsError < StandardError; end
|
10
11
|
|
11
12
|
attr_accessor :max_jobs
|
12
13
|
|
14
|
+
# ## Options
|
15
|
+
# * max_jobs: the maximum number of jobs that can be running at any one time, or nil if unlimited.
|
13
16
|
def initialize(options={})
|
14
17
|
@mutex ||= Mutex.new
|
15
18
|
|
data/lib/job_pool/job.rb
CHANGED
@@ -11,36 +11,64 @@ class JobPool; end
|
|
11
11
|
# A job keeps track of the child process that gets forked.
|
12
12
|
# job is the Ruby data structure, process is the Unix process.
|
13
13
|
class JobPool::Job
|
14
|
-
attr_reader :start_time, :stop_time
|
15
|
-
attr_reader :
|
16
|
-
|
17
|
-
#
|
18
|
-
#
|
19
|
-
#
|
20
|
-
|
14
|
+
attr_reader :start_time, :stop_time # start and finish times of this job
|
15
|
+
attr_reader :stdin, :stdout, :stderr # fds for child's stdin/stdout/stderr
|
16
|
+
|
17
|
+
# **internal**: Use [JobPool#launch], don't call this method directly.
|
18
|
+
#
|
19
|
+
# Starts a process.
|
20
|
+
#
|
21
|
+
# ## Parameters
|
22
|
+
#
|
23
|
+
# * pool [JobPool]: The pool that will contain this job.
|
24
|
+
# * command [String, Array]: The command to run. Can be specified either
|
25
|
+
# as a string or an array of arguments for Process.spawn.
|
26
|
+
#
|
27
|
+
# ## Options
|
28
|
+
#
|
29
|
+
# * stdin [IO, String]: The child's input. If an IO object isn't supplied,
|
30
|
+
# an IOString will be created by calling the parameter's to_s method.
|
31
|
+
# * stdout [IO]: the IO object to receive the child's output.
|
32
|
+
# * stderr [IO]: the IO object to receive the child's stderr.
|
33
|
+
# * timeout [seconds]: the number of seconds to wait before killing the job.
|
34
|
+
#
|
35
|
+
# If `stdin`, `stdout`, or `stderr` are omitted, an empty IOString will be created.
|
36
|
+
# If output and error are IOStrings, the [output] method will return the child's
|
37
|
+
# stdout, and [error] will return its stderr.
|
38
|
+
#
|
39
|
+
# ## Examples
|
40
|
+
#
|
41
|
+
# * Simple invocation: `job = Job.new pool, 'echo hi'`
|
42
|
+
# * Redirect outpout to a file: `Job.new pool, 'wkhtmltopdf', stdout: File.new('/tmp/out.pdf', 'w')`
|
43
|
+
# * Passing an array and options: `Job.new pool, ['cat', '/tmp/infile', {pgroup: true}]`
|
44
|
+
|
45
|
+
def initialize pool, command, options={}
|
21
46
|
@start_time = Time.now
|
22
|
-
@pool
|
23
|
-
@
|
24
|
-
@
|
25
|
-
@outio = outio || StringIO.new
|
26
|
-
@errio = errio || StringIO.new
|
27
|
-
@chin, @chout, @cherr, @child = Open3.popen3(*cmd)
|
47
|
+
@pool = pool
|
48
|
+
@killed = false
|
49
|
+
@timed_out = false
|
28
50
|
|
29
|
-
@
|
51
|
+
@stdin = options[:stdin] || StringIO.new
|
52
|
+
@stdin = StringIO.new(@stdin.to_s) unless @stdin.respond_to?(:readpartial)
|
53
|
+
@stdout = options[:stdout] || StringIO.new
|
54
|
+
@stderr = options[:stderr] || StringIO.new
|
55
|
+
|
56
|
+
@chin, @chout, @cherr, @child = Open3.popen3(*command)
|
30
57
|
@chout.binmode
|
31
58
|
|
32
|
-
@
|
33
|
-
@timed_out = false
|
59
|
+
@pool._add(self)
|
34
60
|
|
35
|
-
@thrin = Thread.new { drain(@
|
36
|
-
@throut = Thread.new { drain(@chout, @
|
37
|
-
@threrr = Thread.new { drain(@cherr, @
|
61
|
+
@thrin = Thread.new { drain(@stdin, @chin) }
|
62
|
+
@throut = Thread.new { drain(@chout, @stdout) }
|
63
|
+
@threrr = Thread.new { drain(@cherr, @stderr) }
|
38
64
|
|
39
65
|
# ensure cleanup is called when the child exits. (crazy that this requires a whole new thread!)
|
40
66
|
@cleanup_thread = Thread.new do
|
41
|
-
if timeout
|
42
|
-
|
43
|
-
|
67
|
+
if options[:timeout]
|
68
|
+
unless @child.join(timeout)
|
69
|
+
@timed_out = true
|
70
|
+
kill
|
71
|
+
end
|
44
72
|
else
|
45
73
|
@child.join
|
46
74
|
end
|
@@ -48,20 +76,25 @@ class JobPool::Job
|
|
48
76
|
end
|
49
77
|
end
|
50
78
|
|
79
|
+
# @param [Hash] opts the options to create a message with.
|
80
|
+
# @option opts [String] :subject The subject
|
81
|
+
# @option opts [String] :from ('nobody') From address
|
82
|
+
# @option opts [String] :to Recipient email
|
83
|
+
# @option opts [String] :body ('') The email's body
|
51
84
|
def write *args
|
52
|
-
@
|
85
|
+
@stdin.write *args
|
53
86
|
end
|
54
87
|
|
55
88
|
def read *args
|
56
|
-
@
|
89
|
+
@stdout.read *args
|
57
90
|
end
|
58
91
|
|
59
92
|
def output
|
60
|
-
@
|
93
|
+
@stdout.string
|
61
94
|
end
|
62
95
|
|
63
96
|
def error
|
64
|
-
@
|
97
|
+
@stderr.string
|
65
98
|
end
|
66
99
|
|
67
100
|
def finished?
|
@@ -131,11 +164,6 @@ private
|
|
131
164
|
@cleanup_thread.join unless Thread.current == @cleanup_thread
|
132
165
|
end
|
133
166
|
|
134
|
-
def outatime
|
135
|
-
@timed_out = true
|
136
|
-
kill
|
137
|
-
end
|
138
|
-
|
139
167
|
# reads every last drop, then closes both files. must be threadsafe.
|
140
168
|
def drain reader, writer
|
141
169
|
begin
|
data/lib/job_pool/version.rb
CHANGED
data/spec/job_pool/job_spec.rb
CHANGED
@@ -5,7 +5,7 @@ require 'job_pool/job'
|
|
5
5
|
# while rspec spec/job_pool/job_spec.rb ; do : ; done
|
6
6
|
|
7
7
|
describe JobPool::Job do
|
8
|
-
class FakeJobPool
|
8
|
+
class FakeJobPool # :nodoc: massively oversimplified job pool used in some testing
|
9
9
|
def initialize
|
10
10
|
@jobs = []
|
11
11
|
end
|
@@ -24,11 +24,9 @@ describe JobPool::Job do
|
|
24
24
|
end
|
25
25
|
|
26
26
|
let(:pool) { FakeJobPool.new }
|
27
|
-
let(:
|
28
|
-
let(:chout) { StringIO.new }
|
29
|
-
let(:cherr) { StringIO.new }
|
27
|
+
let(:small_input) { StringIO.new('small instring') }
|
30
28
|
|
31
|
-
def time_this_block &block
|
29
|
+
def time_this_block &block # :nodoc:
|
32
30
|
start = Time.now
|
33
31
|
block.call
|
34
32
|
finish = Time.now
|
@@ -38,9 +36,9 @@ describe JobPool::Job do
|
|
38
36
|
|
39
37
|
it "has a working drain method" do
|
40
38
|
bigin = StringIO.new('x' * 1024 * 1024) # at least 1 MB of data to test drain loop
|
41
|
-
job = JobPool::Job.new(pool, 'cat', bigin
|
39
|
+
job = JobPool::Job.new(pool, 'cat', stdin: bigin)
|
42
40
|
job.stop
|
43
|
-
expect(
|
41
|
+
expect(job.output).to eq bigin.string
|
44
42
|
expect(job.finished?).to eq true
|
45
43
|
end
|
46
44
|
|
@@ -48,19 +46,20 @@ describe JobPool::Job do
|
|
48
46
|
# pile a bunch of checks into this test so we only have to sleep once
|
49
47
|
expect(pool.count).to eq 0
|
50
48
|
claimed = nil
|
49
|
+
job = nil
|
51
50
|
|
52
51
|
elapsed = time_this_block do
|
53
52
|
# echo -n doesn't work here because of platform variations
|
54
53
|
# and for some reason jruby requires the explicit subshell; mri launches it automatically
|
55
|
-
|
54
|
+
job = JobPool::Job.new(pool, '/bin/sh -c "sleep 0.1 && printf done."', stdin: small_input)
|
56
55
|
expect(pool.count).to eq 1
|
57
|
-
|
58
|
-
expect(
|
59
|
-
expect(
|
60
|
-
claimed =
|
61
|
-
expect(
|
62
|
-
expect(
|
63
|
-
expect(
|
56
|
+
job.stop
|
57
|
+
expect(job.start_time).not_to eq nil
|
58
|
+
expect(job.stop_time).not_to eq nil
|
59
|
+
claimed = job.stop_time - job.start_time
|
60
|
+
expect(job.output).to eq 'done.'
|
61
|
+
expect(job.finished?).to eq true
|
62
|
+
expect(job.success?).to eq true
|
64
63
|
end
|
65
64
|
|
66
65
|
# ensure process elapsed time is in the ballpark
|
@@ -69,36 +68,37 @@ describe JobPool::Job do
|
|
69
68
|
expect(claimed).to be <= elapsed
|
70
69
|
|
71
70
|
expect(pool.count).to eq 0
|
72
|
-
expect(
|
73
|
-
expect(
|
71
|
+
expect(job.stdout.closed_read?).to eq true
|
72
|
+
expect(job.stderr.closed_read?).to eq true
|
74
73
|
end
|
75
74
|
|
76
75
|
it "has a working kill method" do
|
76
|
+
job = nil
|
77
77
|
elapsed = time_this_block do
|
78
|
-
|
78
|
+
job = JobPool::Job.new(pool, ['sleep', '0.5'], stdin: small_input)
|
79
79
|
|
80
|
-
expect(
|
81
|
-
expect(
|
82
|
-
expect(
|
83
|
-
expect(
|
80
|
+
expect(job.finished?).to eq false
|
81
|
+
expect(job.killed?).to eq false
|
82
|
+
expect(job.success?).to eq false
|
83
|
+
expect(job.timed_out?).to eq false
|
84
84
|
|
85
|
-
|
85
|
+
job.kill
|
86
86
|
|
87
|
-
expect(
|
88
|
-
expect(
|
89
|
-
expect(
|
90
|
-
expect(
|
87
|
+
expect(job.finished?).to eq true
|
88
|
+
expect(job.killed?).to eq true
|
89
|
+
expect(job.success?).to eq false
|
90
|
+
expect(job.timed_out?).to eq false
|
91
91
|
end
|
92
92
|
|
93
93
|
expect(elapsed).to be < 0.5
|
94
|
-
expect(
|
95
|
-
expect(
|
94
|
+
expect(job.stdout.closed_read?).to eq true
|
95
|
+
expect(job.stderr.closed_read?).to eq true
|
96
96
|
end
|
97
97
|
|
98
98
|
it "handles invalid commands" do
|
99
99
|
expect {
|
100
100
|
expect(pool.count).to eq 0
|
101
|
-
|
101
|
+
job = JobPool::Job.new(pool, ['ThisCmdDoes.Not.Exist.'], stdin: small_input)
|
102
102
|
raise "we shouldn't get here"
|
103
103
|
}.to raise_error(/[Nn]o such file/)
|
104
104
|
expect(pool.count).to eq 0
|
@@ -106,15 +106,14 @@ describe JobPool::Job do
|
|
106
106
|
|
107
107
|
it "has a working timeout" do
|
108
108
|
elapsed = time_this_block do
|
109
|
-
|
109
|
+
job = JobPool::Job.new(pool, ['sleep', '10'], stdin: small_input, timeout: 0.1)
|
110
110
|
end
|
111
111
|
expect(elapsed).to be < 0.2
|
112
112
|
end
|
113
113
|
|
114
|
-
# TODO: should probably define exactly what happens in this case
|
115
114
|
it "accepts a 0-length timeout" do
|
116
115
|
elapsed = time_this_block do
|
117
|
-
|
116
|
+
job = JobPool::Job.new(pool, ['sleep', '10'], stdin: small_input, timeout: 0)
|
118
117
|
end
|
119
118
|
expect(elapsed).to be < 0.2
|
120
119
|
end
|
data/spec/job_pool_spec.rb
CHANGED
@@ -29,10 +29,10 @@ describe JobPool do
|
|
29
29
|
after { expect(pool.count).to eq 0 }
|
30
30
|
|
31
31
|
it "counts and kills multiple processes" do
|
32
|
-
pool.launch(['sleep', '20']
|
33
|
-
pool.launch(['sleep', '20']
|
34
|
-
pool.launch(['sleep', '20']
|
35
|
-
pool.launch(['sleep', '20']
|
32
|
+
pool.launch(['sleep', '20'])
|
33
|
+
pool.launch(['sleep', '20'])
|
34
|
+
pool.launch(['sleep', '20'])
|
35
|
+
pool.launch(['sleep', '20'])
|
36
36
|
expect(pool.count).to eq 4
|
37
37
|
pool.first.kill
|
38
38
|
expect(pool.count).to eq 3
|
@@ -43,9 +43,9 @@ describe JobPool do
|
|
43
43
|
it "waits for multiple processes" do
|
44
44
|
# these sleep durations might be too small, depends on machine load and scheduling.
|
45
45
|
# if you're seeing threads finishing in the wrong order, try increasing them 10X.
|
46
|
-
process1 = pool.launch(['sleep', '.3']
|
47
|
-
process2 = pool.launch(['sleep', '.1']
|
48
|
-
process3 = pool.launch(['sleep', '.2']
|
46
|
+
process1 = pool.launch(['sleep', '.3'])
|
47
|
+
process2 = pool.launch(['sleep', '.1'])
|
48
|
+
process3 = pool.launch(['sleep', '.2'])
|
49
49
|
expect(pool.count).to eq 3
|
50
50
|
|
51
51
|
child = pool.wait_next
|
@@ -65,7 +65,8 @@ describe JobPool do
|
|
65
65
|
it "handles waiting for zero processes" do
|
66
66
|
expect {
|
67
67
|
child = pool.wait_next
|
68
|
-
|
68
|
+
# if I don't use a string, rdoc claims ThreadsWait is my class. Bug?
|
69
|
+
}.to raise_exception(Object.const_get 'ThreadsWait::ErrNoWaitingThread')
|
69
70
|
end
|
70
71
|
end
|
71
72
|
|
data/spec/readme_spec.rb
CHANGED
@@ -4,7 +4,7 @@ require 'job_pool'
|
|
4
4
|
describe 'README' do
|
5
5
|
it "can do the first example" do
|
6
6
|
pool = JobPool.new
|
7
|
-
job = pool.launch("sleep 0.1; tr A-Za-z N-ZA-Mn-za-m", "the secrets")
|
7
|
+
job = pool.launch("sleep 0.1; tr A-Za-z N-ZA-Mn-za-m", stdin: "the secrets")
|
8
8
|
expect(job.output).to eq ''
|
9
9
|
expect(pool.count).to eq 1
|
10
10
|
sleep(0.2)
|
@@ -16,10 +16,12 @@ describe 'README' do
|
|
16
16
|
pool = JobPool.new
|
17
17
|
# can't use `expect { ... }.to output('contents').to_stdout`
|
18
18
|
# because the test's stdout gets closed
|
19
|
-
|
20
|
-
|
19
|
+
source = File.open('spec/contents.txt.gz')
|
20
|
+
destination = File.open("/tmp/test-#{$$}-out.txt", 'w')
|
21
|
+
pool.launch 'gunzip --to-stdout', stdin: source, stdout: destination
|
21
22
|
pool.wait_next
|
22
|
-
expect(
|
23
|
+
expect(File.read "/tmp/test-#{$$}-out.txt").to eq "contents\n"
|
24
|
+
File.delete "/tmp/test-#{$$}-out.txt"
|
23
25
|
end
|
24
26
|
|
25
27
|
it "can do the killer example" do
|
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: job_pool
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: '0.
|
4
|
+
version: '0.6'
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Scott Bronson
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2015-10-
|
11
|
+
date: 2015-10-12 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: bundler
|
@@ -104,3 +104,4 @@ test_files:
|
|
104
104
|
- spec/job_pool_spec.rb
|
105
105
|
- spec/readme_spec.rb
|
106
106
|
- spec/spec_helper.rb
|
107
|
+
has_rdoc:
|