arvados-cli 0.1.20140403152751 → 0.1.20140403155249

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 1790f4c0faccd8bca4580c3f0e7d1024c849ac2a
-  data.tar.gz: b3f82ba4927f1e9600724ff04e8468bf90c98082
+  metadata.gz: 5ae48eceb6e1a15ee8b14c99aeb18680a103fe28
+  data.tar.gz: 0ff02e615a2a053a80a9671259e474854fdee58e
 SHA512:
-  metadata.gz: 84eeff7d3b99305b384e7d1cac3db1de5468dffb4168ee2e1a49ffbcbd47c688a9c93a70673f16bfb4304214fddddd66012b1ea8facc7b5d12900f7d08240a68
-  data.tar.gz: 2ea3ac5c92980197cec8b5cf4f8a45a9611a9a3d1f05640d2f12b2036c8d4081ab63160e27c59916d3a093b1663c4164c382238056d924d362e7e10daa25a1dd
+  metadata.gz: 33467b027edfdd6ef6ffb138b0a0868e6bc6a0b7f54909bd09022d10fa4f9f5900d4e70c26360fbcf5fe2e40aa6551165fe12c13d70714c2f3fcbc55017dca4c
+  data.tar.gz: 39fcee1939a25c2522bc27964676f55324a7b3579a702547740a392628b950a37dae06e0843e50dba1423a27c9705f61a7c302cf3158e47f5c67b7d811688e01

data/bin/crunch-job

@@ -77,9 +77,9 @@ use POSIX ':sys_wait_h';
 use Fcntl qw(F_GETFL F_SETFL O_NONBLOCK);
 use Arvados;
 use Getopt::Long;
-use Warehouse;
-use Warehouse::Stream;
-use IPC::System::Simple qw(capturex);
+use IPC::Open2;
+use IO::Select;
+use File::Temp;
 use Fcntl ':flock';
 
 $ENV{"TMPDIR"} ||= "/tmp";
@@ -175,11 +175,8 @@ else
 }
 $job_id = $Job->{'uuid'};
 
-$metastream = Warehouse::Stream->new(whc => new Warehouse);
-$metastream->clear;
-$metastream->name('.');
-$metastream->write_start($job_id . '.log.txt');
-
+my $keep_logfile = $job_id . '.log.txt';
+my $local_logfile = File::Temp->new();
 
 $Job->{'runtime_constraints'} ||= {};
 $Job->{'runtime_constraints'}->{'max_tasks_per_node'} ||= 0;
@@ -761,7 +758,7 @@ if ($job_has_uuid) {
 if ($Job->{'output'})
 {
   eval {
-    my $manifest_text = capturex("whget", $Job->{'output'});
+    my $manifest_text = `arv keep get \Q$Job->{'output'}\E`;
     $arv->{'collections'}->{'create'}->execute('collection' => {
       'uuid' => $Job->{'output'},
       'manifest_text' => $manifest_text,
@@ -1071,12 +1068,24 @@ sub process_stderr
   } split ("\n", $jobstep[$job]->{stderr});
 }
 
+sub fetch_block
+{
+  my $hash = shift;
+  my ($keep, $child_out, $output_block);
+
+  my $cmd = "arv keep get \Q$hash\E";
+  open($keep, '-|', $cmd) or die "fetch_block: $cmd: $!";
+  sysread($keep, $output_block, 64 * 1024 * 1024);
+  close $keep;
+  return $output_block;
+}
 
 sub collate_output
 {
-  my $whc = Warehouse->new;
   Log (undef, "collate");
-  $whc->write_start (1);
+
+  my ($child_out, $child_in);
+  my $pid = open2($child_out, $child_in, 'arv', 'keep', 'put', '--raw');
   my $joboutput;
   for (@jobstep)
   {
@@ -1087,26 +1096,36 @@ sub collate_output
     if ($output !~ /^[0-9a-f]{32}(\+\S+)*$/)
     {
       $output_in_keep ||= $output =~ / [0-9a-f]{32}\S*\+K/;
-      $whc->write_data ($output);
+      print $child_in $output;
     }
     elsif (@jobstep == 1)
     {
       $joboutput = $output;
-      $whc->write_finish;
+      last;
     }
-    elsif (defined (my $outblock = $whc->fetch_block ($output)))
+    elsif (defined (my $outblock = fetch_block ($output)))
     {
       $output_in_keep ||= $outblock =~ / [0-9a-f]{32}\S*\+K/;
-      $whc->write_data ($outblock);
+      print $child_in $outblock;
     }
     else
     {
-      my $errstr = $whc->errstr;
-      $whc->write_data ("XXX fetch_block($output) failed: $errstr XXX\n");
+      Log (undef, "XXX fetch_block($output) failed XXX");
       $main::success = 0;
     }
   }
-  $joboutput = $whc->write_finish if !defined $joboutput;
+  $child_in->close;
+
+  if (!defined $joboutput) {
+    my $s = IO::Select->new($child_out);
+    if ($s->can_read(120)) {
+      sysread($child_out, $joboutput, 64 * 1024 * 1024);
+    } else {
+      Log (undef, "timed out reading from 'arv keep put'");
+    }
+  }
+  waitpid($pid, 0);
+
   if ($joboutput)
   {
     Log (undef, "output $joboutput");
@@ -1181,8 +1200,9 @@ sub Log				# ($jobstep_id, $logmessage)
   }
   print STDERR ((-t STDERR) ? ($datetime." ".$message) : $message);
 
-  return if !$metastream;
-  $metastream->write_data ($datetime . " " . $message);
+  if ($metastream) {
+    print $metastream $datetime . " " . $message;
+  }
 }
 
 
@@ -1211,16 +1231,15 @@ sub cleanup
 sub save_meta
 {
   my $justcheckpoint = shift; # false if this will be the last meta saved
-  my $m = $metastream;
-  $m = $m->copy if $justcheckpoint;
-  $m->write_finish;
-  my $whc = Warehouse->new;
-  my $loglocator = $whc->store_block ($m->as_string);
-  $arv->{'collections'}->{'create'}->execute('collection' => {
-    'uuid' => $loglocator,
-    'manifest_text' => $m->as_string,
-  });
-  undef $metastream if !$justcheckpoint; # otherwise Log() will try to use it
+  return if $justcheckpoint;  # checkpointing is not relevant post-Warehouse.pm
+
+  $local_logfile->flush;
+  my $cmd = "arv keep put --filename \Q$keep_logfile\E "
+      . quotemeta($local_logfile->filename);
+  my $loglocator = `$cmd`;
+  die "system $cmd failed: $?" if $?;
+
+  $local_logfile = undef;   # the temp file is automatically deleted
   Log (undef, "log manifest is $loglocator");
   $Job->{'log'} = $loglocator;
   $Job->update_attributes('log', $loglocator) if $job_has_uuid;
@@ -1266,65 +1285,6 @@ sub freeze
 sub thaw
 {
   croak ("Thaw not implemented");
-
-  my $whc;
-  my $key = shift;
-  Log (undef, "thaw from $key");
-
-  @jobstep = ();
-  @jobstep_done = ();
-  @jobstep_todo = ();
-  @jobstep_tomerge = ();
-  $jobstep_tomerge_level = 0;
-  my $frozenjob = {};
-
-  my $stream = new Warehouse::Stream ( whc => $whc,
-				       hash => [split (",", $key)] );
-  $stream->rewind;
-  while (my $dataref = $stream->read_until (undef, "\n\n"))
-  {
-    if ($$dataref =~ /^job /)
-    {
-      foreach (split ("\n", $$dataref))
-      {
-	my ($k, $v) = split ("=", $_, 2);
-	$frozenjob->{$k} = freezeunquote ($v);
-      }
-      next;
-    }
-
-    if ($$dataref =~ /^merge (\d+) (.*)/)
-    {
-      $jobstep_tomerge_level = $1;
-      @jobstep_tomerge
-	  = map { freezeunquote ($_) } split ("\n", freezeunquote($2));
-      next;
-    }
-
-    my $Jobstep = { };
-    foreach (split ("\n", $$dataref))
-    {
-      my ($k, $v) = split ("=", $_, 2);
-      $Jobstep->{$k} = freezeunquote ($v) if $k;
-    }
-    $Jobstep->{'failures'} = 0;
-    push @jobstep, $Jobstep;
-
-    if ($Jobstep->{exitcode} eq "0")
-    {
-      push @jobstep_done, $#jobstep;
-    }
-    else
-    {
-      push @jobstep_todo, $#jobstep;
-    }
-  }
-
-  foreach (qw (script script_version script_parameters))
-  {
-    $Job->{$_} = $frozenjob->{$_};
-  }
-  $Job->save if $job_has_uuid;
 }
 
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: arvados-cli
 version: !ruby/object:Gem::Version
-  version: 0.1.20140403152751
+  version: 0.1.20140403155249
 platform: ruby
 authors:
 - Arvados Authors
@@ -146,7 +146,7 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '0.8'
-description: This is the Arvados SDK CLI gem, git revision 3854a47d632dde711f57c047e80ca84b0b786815
+description: This is the Arvados SDK CLI gem, git revision 39717c4347c00ee0697942d60a50000bfc89fe3c
 email: gem-dev@curoverse.com
 executables:
 - arv