# == Synopsis
#
-# wh-run-pipeline-instance --template pipeline-template-uuid [options] [--] [parameters]
-# wh-run-pipeline-instance --instance pipeline-instance-uuid [options]
+# arv-run-pipeline-instance --template pipeline-template-uuid [options] [--] [parameters]
+# arv-run-pipeline-instance --instance pipeline-instance-uuid [options]
#
# Satisfy a pipeline template by finding or submitting a mapreduce job
# for each pipeline component.
# to finish. Just find out whether jobs are finished,
# queued, or running for each component
#
-# [--create-instance-only] Do not try to satisfy any components. Just
+# [--submit] Do not try to satisfy any components. Just
# create an instance, print its UUID to
# stdout, and exit.
#
# [--no-wait] Make only as much progress as possible without entering
# a sleep/poll loop.
#
-# [--no-reuse-finished] Do not reuse existing outputs to satisfy
-# pipeline components. Always submit a new job
-# or use an existing job which has not yet
-# finished.
-#
# [--no-reuse] Do not reuse existing jobs to satisfy pipeline
# components. Submit a new job for every component.
#
begin
require 'rubygems'
- require 'google/api_client'
require 'json'
require 'pp'
require 'trollop'
-rescue LoadError
+ require 'google/api_client'
+rescue LoadError => l
+ puts $:
abort <<-EOS
-#{$0}: fatal: some runtime dependencies are missing.
+#{$0}: fatal: #{l.message}
+Some runtime dependencies may be missing.
Try: gem install pp google-api-client json trollop
EOS
end
"Do not wait for jobs to finish. Just look up status, submit new jobs if needed, and exit.",
:short => :none,
:type => :boolean)
- opt(:no_reuse_finished,
- "Do not reuse existing outputs to satisfy pipeline components. Always submit a new job or use an existing job which has not yet finished.",
- :short => :none,
- :type => :boolean)
opt(:no_reuse,
"Do not reuse existing jobs to satisfy pipeline components. Submit a new job for every component.",
:short => :none,
"UUID of pipeline instance.",
:short => :none,
:type => :string)
- opt(:create_instance_only,
+ opt(:submit,
"Do not try to satisfy any components. Just create a pipeline instance and output its UUID.",
:short => :none,
:type => :boolean)
+ opt(:run_here,
+ "Manage the pipeline in process.",
+ :short => :none,
+ :type => :boolean)
stop_on [:'--']
end
$options = Trollop::with_standard_exception_handling p do
$debuglevel = $options[:debug_level] || ($options[:debug] && 1) || 0
if $options[:instance]
- if $options[:template] or $options[:create_instance_only]
- abort "#{$0}: syntax error: --instance cannot be combined with --template or --create-instance-only."
+ if $options[:template] or $options[:submit]
+ abort "#{$0}: syntax error: --instance cannot be combined with --template or --submit."
end
elsif not $options[:template]
abort "#{$0}: syntax error: you must supply a --template or --instance."
end
+if $options[:run_here] == $options[:submit]
+ abort "#{$0}: syntax error: you must supply either --run-here or --submit."
+end
+
+# Suppress SSL certificate checks if ARVADOS_API_HOST_INSECURE
+
+module Kernel
+ def suppress_warnings
+ original_verbosity = $VERBOSE
+ $VERBOSE = nil
+ result = yield
+ $VERBOSE = original_verbosity
+ return result
+ end
+end
+
+if ENV['ARVADOS_API_HOST_INSECURE']
+ suppress_warnings { OpenSSL::SSL::VERIFY_PEER = OpenSSL::SSL::VERIFY_NONE }
+end
+
# Set up the API client.
$client ||= Google::APIClient.
if j.is_a? Hash and j[:uuid]
@cache[j[:uuid]] = j
else
- debuglog "create job: #{j[:errors] rescue nil}", 0
+ debuglog "create job: #{j[:errors] rescue nil} with attribute #{attributes}", 0
nil
end
end
moretodo = false
@components.each do |cname, c|
job = nil
- if !c[:job] and
- c[:script_parameters].select { |pname, p| p.is_a? Hash }.empty?
- # Job is fully specified (all parameter values are present) but
- # no particular job has been found.
-
- debuglog "component #{cname} ready to satisfy."
-
- c.delete :wait
- second_place_job = nil # satisfies component, but not finished yet
-
- (@options[:no_reuse] ? [] : JobCache.
- where(script: c[:script],
- script_parameters: c[:script_parameters],
- script_version_descends_from: c[:script_version])
- ).each do |candidate_job|
- candidate_params_downcase = Hash[candidate_job[:script_parameters].
- map { |k,v| [k.downcase,v] }]
- c_params_downcase = Hash[c[:script_parameters].
- map { |k,v| [k.downcase,v] }]
- debuglog "component #{cname} considering job #{candidate_job[:uuid]} version #{candidate_job[:script_version]} parameters #{candidate_params_downcase.inspect}", 3
-
- unless candidate_params_downcase == c_params_downcase
- next
- end
-
- if c[:script_version] !=
- candidate_job[:script_version][0,c[:script_version].length]
- debuglog "component #{cname} would be satisfied by job #{candidate_job[:uuid]} if script_version matched.", 2
- next
- end
-
- unless candidate_job[:success] || candidate_job[:running] ||
- (!candidate_job[:started_at] && !candidate_job[:cancelled_at])
- debuglog "component #{cname} would be satisfied by job #{candidate_job[:uuid]} if it were running or successful.", 2
- next
- end
-
- if candidate_job[:success]
- unless @options[:no_reuse_finished]
- job = candidate_job
- $stderr.puts "using #{job[:uuid]} (finished at #{job[:finished_at]}) for component #{cname}"
- c[:job] = job
- end
- else
- second_place_job ||= candidate_job
- end
- break
- end
- if not c[:job] and second_place_job
- job = second_place_job
- $stderr.puts "using #{job[:uuid]} (running since #{job[:started_at]}) for component #{cname}"
+ if !c[:job] and
+ c[:script_parameters].select { |pname, p| p.is_a? Hash and p[:output_of]}.empty?
+ # No job yet associated with this component and is component inputs
+ # are fully specified (any output_of script_parameters are resolved
+ # to real value)
+ job = JobCache.create({:script => c[:script],
+ :script_parameters => c[:script_parameters],
+ :script_version => c[:script_version],
+ :minimum_script_version => c[:minimum_script_version],
+ :exclude_script_versions => c[:exclude_minimum_script_versions],
+ :nondeterministic => c[:nondeterministic],
+ :no_reuse => @options[:no_reuse]})
+ if job
+ debuglog "component #{cname} new job #{job[:uuid]}"
c[:job] = job
+ else
+ debuglog "component #{cname} new job failed"
end
- if not c[:job]
- debuglog "component #{cname} not satisfied by any existing job."
- if !@options[:dry_run]
- debuglog "component #{cname} new job."
- job = JobCache.create(:script => c[:script],
- :script_parameters => c[:script_parameters],
- :runtime_constraints => c[:runtime_constraints] || {},
- :script_version => c[:script_version] || 'master')
- if job
- debuglog "component #{cname} new job #{job[:uuid]}"
- c[:job] = job
- else
- debuglog "component #{cname} new job failed"
- end
- end
- end
- else
- c[:wait] = true
end
+
if c[:job] and c[:job][:uuid]
if (c[:job][:running] or
not (c[:job][:finished_at] or c[:job][:cancelled_at]))
- c[:job] = JobCache.get(c[:job][:uuid])
+ # Job is running so update copy of job record
+ c[:job] = JobCache.get(c[:job][:uuid])
end
+
if c[:job][:success]
# Populate script_parameters of other components waiting for
# this job
if p.is_a? Hash and p[:output_of] == cname.to_s
debuglog "parameter #{c2name}::#{pname} == #{c[:job][:output]}"
c2[:script_parameters][pname] = c[:job][:output]
+ moretodo = true
end
end
end
elsif c[:job][:running] ||
(!c[:job][:started_at] && !c[:job][:cancelled_at])
- moretodo ||= !@options[:no_wait]
+ # Job is still running
+ moretodo = true
elsif c[:job][:cancelled_at]
debuglog "component #{cname} job #{c[:job][:uuid]} cancelled."
end
@instance[:components] = @components
@instance[:active] = moretodo
report_status
+
+ if @options[:no_wait]
+ moretodo = false
+ end
+
if moretodo
begin
sleep 10
end
end
end
- @instance[:success] = @components.reject { |cname,c| c[:job] and c[:job][:success] }.empty?
+
+ ended = 0
+ succeeded = 0
+ failed = 0
+ @components.each do |cname, c|
+ if c[:job]
+ if c[:job][:finished_at]
+ ended += 1
+ if c[:job][:success] == true
+ succeeded += 1
+ elsif c[:job][:success] == false
+ failed += 1
+ end
+ end
+ end
+ end
+
+ if ended == @components.length or failed > 0
+ @instance[:active] = false
+ @instance[:success] = (succeeded == @components.length)
+ end
+
@instance.save
end
end
runner.apply_parameters(p.leftovers)
runner.setup_instance
- if $options[:create_instance_only]
+ if $options[:submit]
runner.instance.save
puts runner.instance[:uuid]
else
projects / arvados.git / blobdiff