[ARVADOS] updated: e46640c01c65600cbf620cec62a51b0c65cea4ae

git at public.curoverse.com git at public.curoverse.com
Tue Mar 4 17:40:36 EST 2014


Summary of changes:
 .../tutorial-job-debug.html.textile.liquid         |    2 +-
 sdk/cli/bin/crunch-job                             |  197 +++++++++++---------
 2 files changed, 107 insertions(+), 92 deletions(-)

       via  e46640c01c65600cbf620cec62a51b0c65cea4ae (commit)
      from  5aaa5c31fd5d9e00d367cbfc2792fe2335c698b4 (commit)

Those revisions listed above that are new to this repository have
not appeared on any other notification email; so we list those
revisions in full, below.


commit e46640c01c65600cbf620cec62a51b0c65cea4ae
Author: Tim Pierce <twp at curoverse.com>
Date:   Tue Mar 4 17:41:13 2014 -0500

    Remove dependencies on warehouse and IPC::System::Simple (refs #2221).

diff --git a/doc/user/tutorials/tutorial-job-debug.html.textile.liquid b/doc/user/tutorials/tutorial-job-debug.html.textile.liquid
index 38ef3be..560b9ba 100644
--- a/doc/user/tutorials/tutorial-job-debug.html.textile.liquid
+++ b/doc/user/tutorials/tutorial-job-debug.html.textile.liquid
@@ -75,7 +75,7 @@ bc. 2013-12-12_21:36:42 qr1hi-8i9sb-okzukfzkpbrnhst 29827 0 stderr hello world
 The script's output is captured in the log, which is useful for print statement debugging. However, although this script returned a status code of 0 (success), the job failed.  Why?  For a job to complete successfully scripts must explicitly add their output to Keep, and then tell Arvados about it.  Here is a second try:
 
 <notextile>
-<pre><code>~/<b>you</b>/crunch_scripts$ <span class="userinput">cat >hello-world.py <<EOF
+<pre><code>~/<b>you</b>/crunch_scripts$ <span class="userinput">cat >hello-world-fixed.py <<EOF
 #!/usr/bin/env python
 
 import arvados
diff --git a/sdk/cli/bin/crunch-job b/sdk/cli/bin/crunch-job
index e3d0c1d..bd9db89 100755
--- a/sdk/cli/bin/crunch-job
+++ b/sdk/cli/bin/crunch-job
@@ -71,9 +71,8 @@ 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;
 
 $ENV{"TMPDIR"} ||= "/tmp";
 unless (defined $ENV{"CRUNCH_TMP"}) {
@@ -166,10 +165,10 @@ else
 }
 $job_id = $Job->{'uuid'};
 
-$metastream = Warehouse::Stream->new(whc => new Warehouse);
-$metastream->clear;
-$metastream->name('.');
-$metastream->write_start($job_id . '.log.txt');
+# $metastream = Warehouse::Stream->new(whc => new Warehouse);
+# $metastream->clear;
+# $metastream->name('.');
+# $metastream->write_start($job_id . '.log.txt');
 
 
 $Job->{'runtime_constraints'} ||= {};
@@ -738,7 +737,7 @@ if ($job_has_uuid) {
 if ($Job->{'output'})
 {
   eval {
-    my $manifest_text = capturex("whget", $Job->{'output'});
+    my $manifest_text = `arv keep get $Job->{'output'}`;
     $arv->{'collections'}->{'create'}->execute('collection' => {
       'uuid' => $Job->{'output'},
       'manifest_text' => $manifest_text,
@@ -1038,12 +1037,23 @@ sub process_stderr
   } split ("\n", $jobstep[$job]->{stderr});
 }
 
+sub fetch_block
+{
+  my $hash = shift;
+  my ($child_out, $child_in, $output_block);
+
+  my $pid = open2($child_out, $child_in, 'arv', 'keep', 'get', $hash);
+  sysread($child_out, $output_block, 64 * 1024 * 1024);
+  waitpid($pid, 0);
+  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 $joboutput;
   for (@jobstep)
   {
@@ -1054,26 +1064,31 @@ sub collate_output
     if ($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");
+      print $child_in "XXX fetch_block($output) failed XXX\n";
       $main::success = 0;
     }
   }
-  $joboutput = $whc->write_finish if !defined $joboutput;
+  if (!defined $joboutput) {
+    my $s = IO::Select->new($child_out);
+    sysread($child_out, $joboutput, 64 * 1024 * 1024) if $s->can_read(0);
+  }
+  $child_in->close;
+  waitpid($pid, 0);
+
   if ($joboutput)
   {
     Log (undef, "output $joboutput");
@@ -1148,8 +1163,8 @@ sub Log				# ($jobstep_id, $logmessage)
   }
   print STDERR ((-t STDERR) ? ($datetime." ".$message) : $message);
 
-  return if !$metastream;
-  $metastream->write_data ($datetime . " " . $message);
+  # return if !$metastream;
+  # $metastream->write_data ($datetime . " " . $message);
 }
 
 
@@ -1177,20 +1192,20 @@ 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
-  Log (undef, "log manifest is $loglocator");
-  $Job->{'log'} = $loglocator;
-  $Job->update_attributes('log', $loglocator) if $job_has_uuid;
+#  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
+#  Log (undef, "log manifest is $loglocator");
+#  $Job->{'log'} = $loglocator;
+#  $Job->update_attributes('log', $loglocator) if $job_has_uuid;
 }
 
 
@@ -1234,64 +1249,64 @@ 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;
+  # 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;
 }
 
 

-----------------------------------------------------------------------


hooks/post-receive
-- 




More information about the arvados-commits mailing list