Merge branch 'master' into 2895-no-more-redis

[arvados.git] / sdk / cli / bin / crunch-job

diff --git a/sdk/cli/bin/crunch-job b/sdk/cli/bin/crunch-job
index d7d6ae718acec70df8e399dbf5fa1ccd80c6db78..0befdd5dbf9e6c8af770c187824ad90978a05c71 100755 (executable)
--- a/sdk/cli/bin/crunch-job
+++ b/sdk/cli/bin/crunch-job
@@ -76,6 +76,7 @@ use strict;
 use POSIX ':sys_wait_h';
 use Fcntl qw(F_GETFL F_SETFL O_NONBLOCK);
 use Arvados;
+use Digest::MD5 qw(md5_hex);
 use Getopt::Long;
 use IPC::Open2;
 use IO::Select;
@@ -139,7 +140,7 @@ $SIG{'USR2'} = sub
 
 
 my $arv = Arvados->new('apiVersion' => 'v1');
-my $metastream;
+my $local_logfile;
 
 my $User = $arv->{'users'}->{'current'}->execute;
 
@@ -185,7 +186,7 @@ else
 $job_id = $Job->{'uuid'};
 
 my $keep_logfile = $job_id . '.log.txt';
-my $local_logfile = File::Temp->new();
+$local_logfile = File::Temp->new();
 
 $Job->{'runtime_constraints'} ||= {};
 $Job->{'runtime_constraints'}->{'max_tasks_per_node'} ||= 0;
@@ -799,21 +800,37 @@ goto ONELEVEL if !defined $main::success;
 
 release_allocation();
 freeze();
+my $collated_output = &collate_output();
+
 if ($job_has_uuid) {
-  $Job->update_attributes('output' => &collate_output(),
-                          'running' => 0,
-                          'success' => $Job->{'output'} && $main::success,
+  $Job->update_attributes('running' => 0,
+                          'success' => $collated_output && $main::success,
                           'finished_at' => scalar gmtime)
 }
 
-if ($Job->{'output'})
+if ($collated_output)
 {
   eval {
-    my $manifest_text = `arv keep get ''\Q$Job->{'output'}\E`;
-    $arv->{'collections'}->{'create'}->execute('collection' => {
-      'uuid' => $Job->{'output'},
+    open(my $orig_manifest, '-|', 'arv', 'keep', 'get', $collated_output)
+        or die "failed to get collated manifest: $!";
+    # Read the original manifest, and strip permission hints from it,
+    # so we can put the result in a Collection.
+    my @manifest_lines = ();
+    while (my $manifest_line = <$orig_manifest>) {
+      my @words = split(/ /, $manifest_line, -1);
+      foreach my $ii (0..$#words) {
+        if ($words[$ii] =~ /^[0-9a-f]{32}\+/) {
+          $words[$ii] =~ s/\+A[0-9a-f]{40}@[0-9a-f]{8}\b//;
+        }
+      }
+      push(@manifest_lines, join(" ", @words));
+    }
+    my $manifest_text = join("", @manifest_lines);
+    my $output = $arv->{'collections'}->{'create'}->execute('collection' => {
+      'uuid' => md5_hex($manifest_text),
       'manifest_text' => $manifest_text,
     });
+    $Job->update_attributes('output' => $output->{uuid});
     if ($Job->{'output_is_persistent'}) {
       $arv->{'links'}->{'create'}->execute('link' => {
         'tail_kind' => 'arvados#user',
@@ -1251,15 +1268,15 @@ sub Log                         # ($jobstep_id, $logmessage)
   $message =~ s{([^ -\176])}{"\\" . sprintf ("%03o", ord($1))}ge;
   $message .= "\n";
   my $datetime;
-  if ($metastream || -t STDERR) {
+  if ($local_logfile || -t STDERR) {
     my @gmtime = gmtime;
     $datetime = sprintf ("%04d-%02d-%02d_%02d:%02d:%02d",
                         $gmtime[5]+1900, $gmtime[4]+1, @gmtime[3,2,1,0]);
   }
   print STDERR ((-t STDERR) ? ($datetime." ".$message) : $message);
 
-  if ($metastream) {
-    print $metastream $datetime . " " . $message;
+  if ($local_logfile) {
+    print $local_logfile $datetime . " " . $message;
   }
 }
 
@@ -1272,7 +1289,7 @@ sub croak
   freeze() if @jobstep_todo;
   collate_output() if @jobstep_todo;
   cleanup();
-  save_meta() if $metastream;
+  save_meta() if $local_logfile;
   die;
 }
 
@@ -1296,6 +1313,7 @@ sub save_meta
       . quotemeta($local_logfile->filename);
   my $loglocator = `$cmd`;
   die "system $cmd failed: $?" if $?;
+  chomp($loglocator);
 
   $local_logfile = undef;   # the temp file is automatically deleted
   Log (undef, "log manifest is $loglocator");