X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/2c6c1cb11153849e87496bb8d1f5ff24f439a6a4..c6ecb94b3d88dc6c5c28c45e9d602a2abd7f07db:/sdk/cli/bin/crunch-job

diff --git a/sdk/cli/bin/crunch-job b/sdk/cli/bin/crunch-job
index da005a3884..6224a64afe 100755
--- 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;
@@ -500,13 +501,23 @@ if (!$have_slurm)
 
 # If this job requires a Docker image, install that.
 my $docker_bin = "/usr/bin/docker.io";
-my $docker_image = $Job->{runtime_constraints}->{docker_image} || "";
-if ($docker_image) {
+my ($docker_locator, $docker_hash);
+if ($docker_locator = $Job->{docker_image_locator}) {
+  $docker_hash = find_docker_hash($docker_locator);
+  if (!$docker_hash)
+  {
+    croak("No Docker image hash found from locator $docker_locator");
+  }
+  my $docker_install_script = qq{
+if ! $docker_bin images -q --no-trunc | grep -qxF \Q$docker_hash\E; then
+    arv-get \Q$docker_locator/$docker_hash.tar\E | $docker_bin load
+fi
+};
   my $docker_pid = fork();
   if ($docker_pid == 0)
   {
-    srun (["srun", "--nodelist=" . join(' ', @node)],
-          [$docker_bin, 'pull', $docker_image]);
+    srun (["srun", "--nodelist=" . join(',', @node)],
+          ["/bin/sh", "-ec", $docker_install_script]);
     exit ($?);
   }
   while (1)
@@ -515,11 +526,9 @@ if ($docker_image) {
     freeze_if_want_freeze ($docker_pid);
     select (undef, undef, undef, 0.1);
   }
-  # If the Docker image was specified as a hash, pull will fail.
-  # Ignore that error.  We'll see what happens when we try to run later.
-  if (($? != 0) && ($docker_image !~ /^[0-9a-fA-F]{5,64}$/))
+  if ($? != 0)
   {
-    croak("Installing Docker image $docker_image returned exit code $?");
+    croak("Installing Docker image from $docker_locator returned exit code $?");
   }
 }
 
@@ -638,10 +647,10 @@ for (my $todo_ptr = 0; $todo_ptr <= $#jobstep_todo; $todo_ptr ++)
 	  "&& perl -";
     }
     $command .= "&& exec arv-mount --allow-other $ENV{TASK_KEEPMOUNT} --exec ";
-    if ($docker_image)
+    if ($docker_hash)
     {
-      $command .= "crunchstat -cgroup-parent=/sys/fs/cgroup/lxc -cgroup-cid=$ENV{TASK_WORK}/docker.cid ";
-      $command .= "$docker_bin run -i -a stdin -a stdout -a stderr -cidfile=$ENV{TASK_WORK}/docker.cid ";
+      $command .= "crunchstat -cgroup-root=/sys/fs/cgroup -cgroup-parent=docker -cgroup-cid=$ENV{TASK_WORK}/docker.cid -poll=10000 ";
+      $command .= "$docker_bin run -i -a stdin -a stdout -a stderr --cidfile=$ENV{TASK_WORK}/docker.cid ";
       # Dynamically configure the container to use the host system as its
       # DNS server.  Get the host's global addresses from the ip command,
       # and turn them into docker --dns options using gawk.
@@ -654,14 +663,15 @@ for (my $todo_ptr = 0; $todo_ptr <= $#jobstep_todo; $todo_ptr ++)
       }
       while (my ($env_key, $env_val) = each %ENV)
       {
-        if ($env_key =~ /^(JOB|TASK)_/) {
+        if ($env_key =~ /^(ARVADOS|JOB|TASK)_/) {
           $command .= "-e \Q$env_key=$env_val\E ";
         }
       }
-      $command .= "\Q$docker_image\E ";
+      $command .= "\Q$docker_hash\E ";
     } else {
-      $command .= "crunchstat -cgroup-path=/sys/fs/cgroup "
+      $command .= "crunchstat -cgroup-root=/sys/fs/cgroup -poll=10000 "
     }
+    $command .= "stdbuf -o0 -e0 ";
     $command .= "$ENV{CRUNCH_SRC}/crunch_scripts/" . $Job->{"script"};
     my @execargs = ('bash', '-c', $command);
     srun (\@srunargs, \@execargs, undef, $build_script_to_send);
@@ -804,21 +814,39 @@ 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'},
-      'manifest_text' => $manifest_text,
+    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 @stripped_manifest_lines = ();
+    my $orig_manifest_text = '';
+    while (my $manifest_line = <$orig_manifest>) {
+      $orig_manifest_text .= $manifest_line;
+      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(@stripped_manifest_lines, join(" ", @words));
+    }
+    my $stripped_manifest_text = join("", @stripped_manifest_lines);
+    my $output = $arv->{'collections'}->{'create'}->execute('collection' => {
+      'uuid' => md5_hex($stripped_manifest_text),
+      'manifest_text' => $orig_manifest_text,
     });
+    $Job->update_attributes('output' => $output->{uuid});
     if ($Job->{'output_is_persistent'}) {
       $arv->{'links'}->{'create'}->execute('link' => {
         'tail_kind' => 'arvados#user',
@@ -1417,6 +1445,21 @@ sub must_lock_now
   }
 }
 
+sub find_docker_hash {
+  # Given a Keep locator, search for a matching link to find the Docker hash
+  # of the stored image.
+  my $locator = shift;
+  my $links_result = $arv->{links}->{list}->execute(
+    filters => [["head_uuid", "=", $locator],
+                ["link_class", "=", "docker_image_hash"]],
+    limit => 1);
+  my $docker_hash;
+  foreach my $link (@{$links_result->{items}}) {
+    $docker_hash = lc($link->{name});
+  }
+  return $docker_hash;
+}
+
 __DATA__
 #!/usr/bin/perl