7582: Don't call stdbuf in minimal run mode.

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

diff --git a/sdk/cli/bin/crunch-job b/sdk/cli/bin/crunch-job
index 423f4cc043409b0f63f96c6c84333b746e21d19a..8167eda9f0fe527d088fda236cc50fe8ca6c84eb 100755 (executable)
--- a/sdk/cli/bin/crunch-job
+++ b/sdk/cli/bin/crunch-job
@@ -411,7 +411,7 @@ if (!defined $no_clear_tmp) {
 }
 
 # If this job requires a Docker image, install that.
-my ($docker_locator, $docker_stream, $docker_hash, $docker_limitmem);
+my ($docker_locator, $docker_stream, $docker_hash, $docker_limitmem, $dockeruserarg);
 if ($docker_locator = $Job->{docker_image_locator}) {
   ($docker_stream, $docker_hash) = find_docker_image($docker_locator);
   if (!$docker_hash)
@@ -449,6 +449,42 @@ fi
       {fork => 1});
   $docker_limitmem = ($? == 0);
 
+  # Find a non-root Docker user to use.
+  # Tries the default user for the container, then 'crunch', then 'nobody',
+  # testing for whether the actual user id is non-zero.  This defends against
+  # mistakes but not malice, but we intend to harden the security in the future
+  # so we don't want anyone getting used to their jobs running as root in their
+  # Docker containers.
+  my @tryusers = ("", "crunch", "nobody");
+  foreach my $try_user (@tryusers) {
+    my $try_user_arg;
+    if ($try_user eq "") {
+      Log(undef, "Checking if container default user is not UID 0");
+      $try_user_arg = "";
+    } else {
+      Log(undef, "Checking if user '$try_user' is not UID 0");
+      $try_user_arg = "--user=$try_user";
+    }
+    srun(["srun", "--nodelist=" . $node[0]],
+         ["/bin/sh", "-ec",
+          "a=`$docker_bin run --rm --user=$try_user $docker_hash id --user` && " .
+          " test \$a -ne 0"],
+         {fork => 1});
+    if ($? == 0) {
+      $dockeruserarg = $try_user_arg;
+      if ($try_user eq "") {
+        Log(undef, "Container will run with default user");
+      } else {
+        Log(undef, "Container will run with $dockeruserarg");
+      }
+      last;
+    }
+  }
+
+  if (!defined $dockeruserarg) {
+    croak("Could not find a user in container that is not UID 0 (tried default user, @tryusers) or there was a problem running 'id' in the container.");
+  }
+
   if ($Job->{arvados_sdk_version}) {
     # The job also specifies an Arvados SDK version.  Add the SDKs to the
     # tar file for the build script to install.
@@ -859,7 +895,7 @@ for (my $todo_ptr = 0; $todo_ptr <= $#jobstep_todo; $todo_ptr ++)
     {
       my $cidfile = "$ENV{CRUNCH_TMP}/$Jobstep->{arvados_task}->{uuid}-$Jobstep->{failures}.cid";
       $command .= "crunchstat -cgroup-root=/sys/fs/cgroup -cgroup-parent=docker -cgroup-cid=$cidfile -poll=10000 ";
-      $command .= "$docker_bin run --rm=true --attach=stdout --attach=stderr --attach=stdin -i --user=crunch --cidfile=$cidfile --sig-proxy ";
+      $command .= "$docker_bin run --rm=true --attach=stdout --attach=stderr --attach=stdin -i \Q$dockeruserarg\E --cidfile=$cidfile --sig-proxy ";
       # We only set memory limits if Docker lets us limit both memory and swap.
       # Memory limits alone have been supported longer, but subprocesses tend
       # to get SIGKILL if they exceed that without any swap limit set.
@@ -917,8 +953,13 @@ for (my $todo_ptr = 0; $todo_ptr <= $#jobstep_todo; $todo_ptr ++)
       }
       $command .= "--env=\QHOME=$ENV{HOME}\E ";
       $command .= "\Q$docker_hash\E ";
-      $command .= "stdbuf --output=0 --error=0 ";
-      $command .= "perl - $ENV{CRUNCH_SRC}/crunch_scripts/" . $Job->{"script"};
+
+      if ($Job->{arvados_sdk_version}) {
+        $command .= "stdbuf --output=0 --error=0 ";
+        $command .= "perl - \Q$ENV{CRUNCH_SRC}/crunch_scripts/$Job->{script}\E";
+      } else {
+        $command .= "/bin/sh -c \'mkdir -p \"$ENV{JOB_WORK}\" \"$ENV{TASK_WORK}\" && exec \Q$ENV{CRUNCH_SRC}/crunch_scripts/$Job->{script}\E\'";
+      }
     } else {
       # Non-docker run
       $command .= "crunchstat -cgroup-root=/sys/fs/cgroup -poll=10000 ";
@@ -980,7 +1021,7 @@ for (my $todo_ptr = 0; $todo_ptr <= $#jobstep_todo; $todo_ptr ++)
     my $gotsome
        = readfrompipes ()
        + reapchildren ();
-    if (!$gotsome)
+    if (!$gotsome || ($latest_refresh + 2 < scalar time))
     {
       check_refresh_wanted();
       check_squeue();