rbbt-util 5.28.11 → 5.29.2

data/lib/rbbt/workflow/util/orchestrator.rb

@@ -26,15 +26,25 @@ module Workflow
       workload
     end
 
+    def self.workload(jobs)
+      jobs.inject({}) do |acc,job| 
+          Orchestrator.job_workload(job).each do |j,d|
+            acc[j] = d unless acc.keys.collect{|k| k.path }.include? j.path
+          end
+          acc
+        end
+    end
+
     def self.job_rules(rules, job)
       workflow = job.workflow.to_s
       task_name = job.task_name.to_s
+      defaults = rules["defaults"] || {}
 
-      return IndiferentHash.setup(rules["defaults"]) unless rules[workflow]
-      return IndiferentHash.setup(rules["defaults"]) unless rules[workflow][task_name]
+      return IndiferentHash.setup(defaults) unless rules[workflow]
+      return IndiferentHash.setup(defaults) unless rules[workflow][task_name]
 
       job_rules = IndiferentHash.setup(rules[workflow][task_name])
-      rules["defaults"].each{|k,v| job_rules[k] = v if job_rules[k].nil? } if rules["defaults"]
+      defaults.each{|k,v| job_rules[k] = v if job_rules[k].nil? } if defaults
       job_rules
     end
 
@@ -169,12 +179,7 @@ module Workflow
     def process(rules, jobs)
       begin
 
-        workload = jobs.inject({}) do |acc,job| 
-          Orchestrator.job_workload(job).each do |j,d|
-            acc[j] = d unless acc.keys.collect{|k| k.path }.include? j.path
-          end
-          acc
-        end
+        workload = Orchestrator.workload(jobs)
         all_jobs = workload.keys
 
         top_level_jobs = jobs.collect{|job| job.path }

data/lib/rbbt/workflow/util/provenance.rb

@@ -22,11 +22,14 @@ class Step
   end
 
   def self.prov_report_msg(status, name, path, info = nil)
-    parts = path.sub(/\{.*/,'').sub(/#{Regexp.quote(name)}$/,'').split "/"
+    parts = path.sub(/\{.*/,'').split "/"
 
+    parts.pop
+    
     task = Log.color(:yellow, parts.pop)
     workflow = Log.color(:magenta, parts.pop)
-    if status.to_s == 'noinfo' and parts.last != 'jobs'
+    #if status.to_s == 'noinfo' && parts.last != 'jobs'
+    if ! Workflow.job_path?(path)
       task, status, workflow = Log.color(:yellow, info[:task_name]), Log.color(:green, "file"), Log.color(:magenta, "-")
     end
 

data/share/rbbt_commands/slurm/list

@@ -0,0 +1,141 @@
+#!/usr/bin/env ruby
+
+require 'rbbt-util'
+require 'rbbt/util/simpleopt'
+
+#$0 = "rbbt #{$previous_commands*""} #{ File.basename(__FILE__) }" if $previous_commands
+
+options = SOPT.setup <<EOF
+
+Queue a job in Marenostrum
+
+$ rbbt mnl [options] 
+
+-h--help Print this help
+-d--done Done jobs only
+-e--error Error jobs only
+-a--aborted SLURM aboted jobs
+-r--running Running jobs only
+-q--queued Queued jobs only
+-j--job* Job ids
+-s--search* Regular expression
+-t--tail* Show the last lines of the STDERR
+EOF
+
+if options[:help]
+  if defined? rbbt_usage
+    rbbt_usage 
+  else
+    puts SOPT.doc
+  end
+  exit 0
+end
+
+Log.severity = 4
+done, error, running, queued, aborted, jobid, search, tail = options.values_at :done, :error, :running, :queued, :aborted, :job, :search, :tail
+
+workdir = File.expand_path('~/rbbt-slurm')
+Path.setup(workdir)
+
+running_jobs = begin
+                   CMD.cmd('squeue').read.split("\n").collect{|l| l.to_i.to_s}
+               rescue
+                   Log.warn "Cannot determine if jobs are running, they will seem to be all alive (Job ID in green)"
+                   $norunningjobs = true
+                   []
+               end
+
+count = 0
+workdir.glob("**/command.slurm").sort_by{|f| File.mtime(f)}.each do |fcmd|
+    dir = File.dirname(fcmd)
+
+    if m = Open.read(fcmd).match(/#CMD: (.*)/)
+        cmd = m[1]
+    else
+        cmd = nil
+    end
+
+    if m = Open.read(fcmd).match(/# Run command\n(.*?)\n/im)
+        exe = m[1]
+    else
+        exe = nil
+    end
+
+    if m = Open.read(fcmd).match(/^CONTAINER_DIR=(.*)/)
+        container_home = m[1]
+    else
+        container_home = nil
+    end
+
+
+    if File.exists?(fid = File.join(dir, 'job.id'))
+        id = Open.read(fid).chomp
+    else
+        id = nil
+    end
+
+    if File.exists?(fstatus = File.join(dir, 'exit.status'))
+        exit_status = Open.read(fstatus).to_i
+    else
+        exit_status = nil
+    end
+    
+    if File.exists?(fstatus = File.join(dir, 'job.status'))
+        nodes = Open.read(fstatus).split("\n").last.split(/\s+/).last.split(",")
+    else
+        nodes = []
+    end
+
+    if File.exists?(File.join(dir, 'std.out'))
+        outt = File.mtime File.join(dir, 'std.out')
+        errt = File.mtime File.join(dir, 'std.err')
+        time_diff = Time.now - [outt, errt].max
+    end
+
+    fdep = File.join(dir, 'dependencies.list')
+    deps = Open.read(fdep).split("\n") if File.exists?(fdep)
+
+    if done || error || aborted || running || queued || jobid || search
+      select = false
+      select = true if done && exit_status == 0
+      select = true if error && exit_status && exit_status != 0 
+      select = true if aborted && (exit_status.nil? && ! running_jobs.include?(id))
+      select = true if queued && deps && (running_jobs & deps).any?
+      select = true if running && (exit_status.nil? && running_jobs.include?(id)) && (!deps || (running_jobs & deps).empty?)
+      select = true if jobid && jobid.split(",").include?(id)
+      select = true if search && cmd.match(/#{search}/)
+      next unless select 
+    end
+
+
+    puts Log.color :blue, dir
+    puts Log.color(:magenta, "Creation: ") << File.mtime(File.join(dir, 'command.slurm')).to_s
+    puts Log.color(:magenta, "Done: ") << File.mtime(File.join(dir, 'exit.status')).to_s if File.exist?(File.join(dir, 'exit.status'))
+    puts Log.color(:magenta, "Exec: ") << (exe || "Missing")
+    puts Log.color(:magenta, "CMD: ") << (Log.color(:yellow, cmd) || "Missing")
+    puts Log.color(:magenta, "HOME: ") << Log.color(:yellow, container_home) if container_home
+    puts Log.color(:magenta, "Job ID: ") << (exit_status ? (exit_status == 0 ? Log.color(:green, "Done") : Log.color(:red, "Error")) :  (running_jobs.include?(id) || $norunningjobs ? Log.color(:green, id) : Log.color(:red, id) ))
+    puts Log.color(:magenta, "Dependencies: ") << deps * ", " if deps
+    puts Log.color(:magenta, "Nodes: ") << nodes * ", "
+    puts Log.color(:magenta, "Output: ") << File.exists?(File.join(dir, 'std.out')).to_s << (id.nil? ? "" : " (last update " + Misc.format_seconds(time_diff) + " ago)")
+
+    if tail && File.exists?(File.join(dir, 'std.err'))
+        if exit_status && exit_status != 0
+            puts Log.color(:magenta, "First error or exception found: ")
+            puts CMD.cmd("grep -i -w 'error\\|[a-z]*exception' #{File.join(dir, 'std.err')} -A #{tail.to_i} |head -n #{tail.to_i}", :no_fail => true).read
+        elsif exit_status
+            puts Log.color(:magenta, "Completed jobs: ")
+            puts CMD.cmd("grep -i -w 'Completed step' #{File.join(dir, 'std.err')} | grep -v 'Retrying dep.' | tail -n #{tail.to_i}", :no_fail => true).read
+        else
+            puts Log.color(:magenta, "Log tail: ")
+            puts CMD.cmd("tail -n #{tail.to_i} #{File.join(dir, 'std.err')}").read
+        end
+    end
+
+    count += 1
+
+end
+
+puts 
+puts "Found #{count} jobs"
+

data/share/rbbt_commands/slurm/orchestrate

@@ -0,0 +1,47 @@
+#!/usr/bin/env ruby
+
+require 'rbbt/util/simpleopt'
+require 'rbbt/workflow'
+require 'rbbt/workflow/usage'
+require 'rbbt/hpc'
+require 'rbbt/hpc/orchestrate'
+require 'time'
+
+$slurm_options = SOPT.get <<EOF
+-dr--dry_run Print only the template
+-cj--clean_job Clean job
+--drbbt* Use development version of rbbt 
+-sing--singularity Use Singularity
+-ug--user_group* Use alternative user group for group project directory
+-c--contain* Contain in directory (using Singularity)
+-s--sync* Contain in directory and sync jobs
+-e--exclusive Make exclusive use of the node
+-hm--highmem Make use of highmem cores
+-wc--wipe_container* Wipe the jobs from the contain directory
+-CS--contain_and_sync Contain and sync to default locations
+-ci--copy_image When using a container directory, copy image there
+-t--tail Tail the logs
+-q--queue* Queue
+-t--task_cpus* Tasks
+-W--workflows* Additional workflows
+-tm--time* Time
+-R--orchestration_rules* Orchestration rules
+-rmb--remove_slurm_basedir Remove the SLURM working directory (command, STDIN, exit status, ...)
+EOF
+
+class Step
+    def run(*args)
+        if done?
+            self.load
+        else
+            begin
+                Log.debug "Issuing SLURM job for #{self.path}"
+                HPC::SLURM.orchestrate_job(self, SOPT::GOT_OPTIONS.merge($slurm_options)) 
+            rescue HPC::SBATCH
+            end
+        end
+    end
+end
+
+ARGV.concat ["-W", $slurm_options[:workflows]] if $slurm_options[:workflows]
+load Rbbt.share.rbbt_commands.workflow.task.find

data/share/rbbt_commands/{workflow/slurm → slurm/task}

@@ -9,8 +9,9 @@ require 'time'
 $slurm_options = SOPT.get <<EOF
 -dr--dry_run Print only the template
 -cj--clean_job Clean job
---drbbt Use development version of rbbt 
+--drbbt* Use development version of rbbt 
 -sing--singularity Use Singularity
+-ug--user_group* Use alternative user group for group project directory
 -c--contain* Contain in directory (using Singularity)
 -s--sync* Contain in directory and sync jobs
 -e--exclusive Make exclusive use of the node
@@ -21,8 +22,9 @@ $slurm_options = SOPT.get <<EOF
 -t--tail Tail the logs
 -q--queue* Queue
 -t--task_cpus* Tasks
+-W--workflows* Additional workflows
 -tm--time* Time
--S--server* SLURM login node
+-rmb--remove_slurm_basedir Remove the SLURM working directory (command, STDIN, exit status, ...)
 EOF
 
 class Step
@@ -30,9 +32,14 @@ class Step
         if done?
             self.load
         else
-            Marenostrum::SLURM.run_job(self, SOPT::GOT_OPTIONS.merge($slurm_options)) 
+            begin
+                Log.debug "Issuing SLURM job for #{self.path}"
+                HPC::SLURM.run_job(self, SOPT::GOT_OPTIONS.merge($slurm_options)) 
+            rescue HPC::SBATCH
+            end
         end
     end
 end
 
+ARGV.concat ["-W", $slurm_options[:workflows]] if $slurm_options[:workflows]
 load Rbbt.share.rbbt_commands.workflow.task.find

data/share/rbbt_commands/workflow/info

@@ -17,6 +17,7 @@ $ rbbt workflow info <job-result>
 -a--all Print all info entries
 -r--recursive Print recursive input values
 -o--original Print original object
+-w--width* Screen width
 EOF
 
 SOPT.usage if options[:help]
@@ -24,6 +25,7 @@ SOPT.usage if options[:help]
 file = ARGV.shift
 all = options.delete :all
 recursive = options.delete :recursive
+width = (options.delete(:width) || 80).to_i
 
 def get_step(file)
   file = file.sub(/\.(info|files)/,'')
@@ -86,6 +88,7 @@ pid = info[:pid]
 exception = info[:exception]
 rest = info.keys - [:inputs, :dependencies, :status, :time_elapsed, :messages, :backtrace, :exception, :pid, :archived_info]
 
+
 puts Log.color(:magenta, "File") << ": " << step.path
 puts Log.color(:magenta, "Status") << ": " << status_msg(status) << ((step.aborted? || step.error?) && step.recoverable_error? ? " (recoverable)" : "" ) << (step.dirty? ? " (dirty)" : "")
 puts Log.color(:magenta, "Pid") << ": " << pid_msg(pid, status.to_s == "done")
@@ -96,14 +99,14 @@ if inputs and inputs.any?
   inputs.each do |input,value|
     case value
     when nil
-      puts Misc.format_definition_list_item("  " + input.to_s, 'nil', 80, 20, :blue)
+      puts Misc.format_definition_list_item("  " + input.to_s, 'nil', width, 20, :blue)
     when Array
-      puts Misc.format_definition_list_item("  " + input.to_s, (value.length > 6 ? value[0..5]*"\n" << "\n" << "..." : value * "\n" ), 80, 20, :blue)
+      puts Misc.format_definition_list_item("  " + input.to_s, (value.length > 6 ? value[0..5]*"\n" << "\n" << "..." : value * "\n" ), width, 20, :blue)
     when TrueClass, FalseClass
-      puts Misc.format_definition_list_item("  " + input.to_s, value.to_s, 80, 20, :blue)
+      puts Misc.format_definition_list_item("  " + input.to_s, value.to_s, width, 20, :blue)
     else
       text = value.to_s.split("\n")[0..5].compact * "\n\n"
-      puts Misc.format_definition_list_item("  " + input.to_s, text, 80, 20, :blue)
+      puts Misc.format_definition_list_item("  " + input.to_s, text, width, 20, :blue)
     end
   end
 end
@@ -162,16 +165,16 @@ if recursive
     inputs.each do |input,value|
       case value
       when nil
-        puts Misc.format_definition_list_item("  " << input.to_s, 'nil', 80, 20, :blue)
+        puts Misc.format_definition_list_item("  " << input.to_s, 'nil', width, 20, :blue)
       when Array
-        puts Misc.format_definition_list_item("  " << input.to_s, (value.length > 6 ? (value[0..5])*"\n\n" << "\n\n" << "..." : value * "\n\n" ), 80, 20, :blue).gsub("\n\n","\n")
+        puts Misc.format_definition_list_item("  " << input.to_s, (value.length > 6 ? (value[0..5])*"\n\n" << "\n\n" << "..." : value * "\n\n" ), width, 20, :blue).gsub("\n\n","\n")
       when TrueClass, FalseClass
-        puts Misc.format_definition_list_item("  " << input.to_s, value.to_s, 80, 20, :blue)
+        puts Misc.format_definition_list_item("  " << input.to_s, value.to_s, width, 20, :blue)
       else
-        lines = value.to_s.split("\n").collect{|l| l.length >= 60 ? l[0..45] + " ..." : l }
+        lines = value.to_s.split("\n").collect{|l| l.length >= width - 5 ? l[0..width - 5] + " ..." : l }
         text = lines[0..5].compact * "\n\n"
         text << "\n\n...\n\n" if lines.length > 6
-        puts Misc.format_definition_list_item("  " << input.to_s, text, 80, 20, :blue).gsub("\n\n","\n")
+        puts Misc.format_definition_list_item("  " << input.to_s, text, width, 20, :blue).gsub("\n\n","\n")
       end
     end
   end

data/test/rbbt/test_workflow.rb

@@ -156,6 +156,15 @@ for this dependency
     Open.read(file).reverse
   end
 
+  task :create_file => :text do |file|
+    Open.write(file('a'), "A")
+    Open.write(file('b'), "B")
+    "DONE"
+  end
+
+  dep_task :reverse_step_file, TestWF, :reverse_file do |jobname, options, dependencies|
+    dep = dependencies.flatten.first
+  end
 
 end
 
@@ -377,9 +386,9 @@ class TestWorkflow < Test::Unit::TestCase
     job.run
     Misc.with_env "RBBT_UPDATE", 'true' do
       assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
-      job = TestWF.job(:t3)
-      job.step(:t1).clean
-      assert job.checks.select{|d| d.task_name.to_s == "t1" }.empty?
+      #job = TestWF.job(:t3)
+      #job.step(:t1).clean
+      #assert job.checks.select{|d| d.task_name.to_s == "t1" }.empty?
       job = TestWF.job(:t3).recursive_clean
       job.run
       assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
@@ -418,7 +427,7 @@ class TestWorkflow < Test::Unit::TestCase
     TmpFile.with_file do |dir|
       Path.setup(dir)
       Step.save_job_inputs(job, dir)
-      assert_equal Dir.glob(dir + "/*"), [dir.file.find + '.read']
+      assert_equal Dir.glob(dir + "/*"), [dir.file.find + '.yaml']
       inputs  = Workflow.load_inputs(dir, [:file], :file => :file)
       assert_equal inputs, {:file => 'code'}
     end
@@ -440,4 +449,27 @@ class TestWorkflow < Test::Unit::TestCase
       end
     end
   end
+
+  def test_input_step_file_check
+    job = TestWF.job(:t3).recursive_clean
+    job.run
+    Misc.with_env "RBBT_UPDATE", 'true' do
+      assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
+      #job = TestWF.job(:t3)
+      #job.step(:t1).clean
+      #assert job.checks.select{|d| d.task_name.to_s == "t1" }.empty?
+      job = TestWF.job(:t3).recursive_clean
+      job.run
+      assert job.checks.select{|d| d.task_name.to_s == "t1" }.any?
+      job = TestWF.job(:t3)
+      sleep 1
+      Open.touch job.step(:t1).path
+      Misc.with_env "RBBT_UPDATE", "false" do
+        assert job.updated?
+      end
+      Misc.with_env "RBBT_UPDATE", "true" do
+        assert ! job.updated?
+      end
+    end
+  end
 end

data/test/rbbt/tsv/test_attach.rb

@@ -566,7 +566,7 @@ row2,CC
       tsv3.keys.each{|k| tsv3[k] = nil if tsv3[k] == ""}
     end
 
-    assert_equal tsv1.attach(tsv2, :complete => true).attach(tsv3, :complete => true)["row1"], [nil, "B", nil]
+    assert_equal [nil, "B", nil], tsv1.attach(tsv2, :complete => true).attach(tsv3, :complete => true)["row1"]
   end
 
   def test_attach_index_both_non_key
@@ -597,10 +597,10 @@ A    Id3
 
     tsv1 = tsv2 = nil
 
-    tsv1 = Rbbt.tmp.test.test1.data.tsv :double,  :sep => /\s+/
-    tsv2 = Rbbt.tmp.test.test2.data.tsv :double,  :sep => /\s+/
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
 
-    tsv2.identifiers = Rbbt.tmp.test.test2.identifiers.produce.find #.to_s
+    tsv2.identifiers = Rbbt.tmp.test.test2.identifiers.produce(true).produce.find #.to_s
 
     tsv1.attach tsv2, :fields => ["ValueE"] #, :persist_input => true
     Log.tsv tsv1
@@ -627,12 +627,92 @@ E    B
 
     tsv1 = tsv2 = nil
 
-    tsv1 = Rbbt.tmp.test.test1.data.tsv :double,  :sep => /\s+/
-    tsv2 = Rbbt.tmp.test.test2.data.tsv :double,  :sep => /\s+/
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
 
     tsv1.attach tsv2, :fields => ["ValueE"] #, :persist_input => true
     Log.tsv tsv1
     
   end
+
+  def test_attach_complete
+    content1 =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueA
+row1    a|aa|aaa
+row2    A
+    EOF
+
+    content2 =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueB
+row1    b
+row3    C
+    EOF
+    Rbbt.claim Rbbt.tmp.test.test1.data, :string, content1
+    Rbbt.claim Rbbt.tmp.test.test2.data, :string, content2
+
+    tsv1 = tsv2 = nil
+
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
+
+    tsv1.attach tsv2, :complete => true
+    assert_equal [[], ["C"]], tsv1["row3"]
+
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
+
+    ppp tsv1.attach tsv2, :complete => ["AA"]
+    tsv1.attach tsv2, :complete => ["AA"]
+    assert_equal [["AA"], ["C"]], tsv1["row3"]
+  end
+
+  def test_attach_complete_identifiers
+    content1 =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueA
+row1    a|aa|aaa
+row2    A
+    EOF
+
+    content2 =<<-EOF
+#: :sep=/\\s+/
+#Id2    ValueB
+ROW_1    b
+ROW_2    C
+    EOF
+
+    identifiers =<<-EOF
+#: :sep=/\\s+/
+#Id    Id2
+row1    ROW_1
+row2    ROW_2
+row3    ROW_3
+    EOF
+    Rbbt.claim Rbbt.tmp.test.test1.data, :string, content1
+    Rbbt.claim Rbbt.tmp.test.test2.data, :string, content2
+    Rbbt.claim Rbbt.tmp.test.identifiers.data, :string, identifiers
+
+    tsv1 = tsv2 = nil
+
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
+    ids = Rbbt.tmp.test.identifiers.data.produce(true).tsv :double,  :sep => /\s+/
+
+    tsv1.identifiers = ids
+
+    tsv1.attach tsv2
+    assert_equal [["A"], ["C"]], tsv1["row2"]
+
+    tsv1 = Rbbt.tmp.test.test1.data.produce(true).tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.produce(true).tsv :double,  :sep => /\s+/
+    ids = Rbbt.tmp.test.identifiers.data.produce(true).tsv :double,  :sep => /\s+/
+
+    tsv1.identifiers = ids
+
+    tsv1.attach tsv2, :complete => true
+    assert_equal [["A"], ["C"]], tsv1["row2"]
+  end
 end