X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/c14246b9a21d038fc6fa850f4032659a98397784..77d9c05d89dabc9e9e9a15f46cd12c8ad61ed64e:/services/api/app/models/container.rb

diff --git a/services/api/app/models/container.rb b/services/api/app/models/container.rb
index 52f1cba723..9420ef3cb8 100644
--- a/services/api/app/models/container.rb
+++ b/services/api/app/models/container.rb
@@ -1,4 +1,5 @@
 require 'whitelist_update'
+require 'safe_json'
 
 class Container < ArvadosModel
   include HasUuid
@@ -81,45 +82,117 @@ class Container < ArvadosModel
     end
   end
 
+  # Create a new container (or find an existing one) to satisfy the
+  # given container request.
+  def self.resolve(req)
+    c_attrs = {
+      command: req.command,
+      cwd: req.cwd,
+      environment: req.environment,
+      output_path: req.output_path,
+      container_image: resolve_container_image(req.container_image),
+      mounts: resolve_mounts(req.mounts),
+      runtime_constraints: resolve_runtime_constraints(req.runtime_constraints),
+      scheduling_parameters: req.scheduling_parameters,
+    }
+    act_as_system_user do
+      if req.use_existing && (reusable = find_reusable(c_attrs))
+        reusable
+      else
+        Container.create!(c_attrs)
+      end
+    end
+  end
+
+  # Return a runtime_constraints hash that complies with requested but
+  # is suitable for saving in a container record, i.e., has specific
+  # values instead of ranges.
+  #
+  # Doing this as a step separate from other resolutions, like "git
+  # revision range to commit hash", makes sense only when there is no
+  # opportunity to reuse an existing container (e.g., container reuse
+  # is not implemented yet, or we have already found that no existing
+  # containers are suitable).
+  def self.resolve_runtime_constraints(runtime_constraints)
+    rc = {}
+    defaults = {
+      'keep_cache_ram' =>
+      Rails.configuration.container_default_keep_cache_ram,
+    }
+    defaults.merge(runtime_constraints).each do |k, v|
+      if v.is_a? Array
+        rc[k] = v[0]
+      else
+        rc[k] = v
+      end
+    end
+    rc
+  end
+
+  # Return a mounts hash suitable for a Container, i.e., with every
+  # readonly collection UUID resolved to a PDH.
+  def self.resolve_mounts(mounts)
+    c_mounts = {}
+    mounts.each do |k, mount|
+      mount = mount.dup
+      c_mounts[k] = mount
+      if mount['kind'] != 'collection'
+        next
+      end
+      if (uuid = mount.delete 'uuid')
+        c = Collection.
+          readable_by(current_user).
+          where(uuid: uuid).
+          select(:portable_data_hash).
+          first
+        if !c
+          raise ArvadosModel::UnresolvableContainerError.new "cannot mount collection #{uuid.inspect}: not found"
+        end
+        if mount['portable_data_hash'].nil?
+          # PDH not supplied by client
+          mount['portable_data_hash'] = c.portable_data_hash
+        elsif mount['portable_data_hash'] != c.portable_data_hash
+          # UUID and PDH supplied by client, but they don't agree
+          raise ArgumentError.new "cannot mount collection #{uuid.inspect}: current portable_data_hash #{c.portable_data_hash.inspect} does not match #{c['portable_data_hash'].inspect} in request"
+        end
+      end
+    end
+    return c_mounts
+  end
+
+  # Return a container_image PDH suitable for a Container.
+  def self.resolve_container_image(container_image)
+    coll = Collection.for_latest_docker_image(container_image)
+    if !coll
+      raise ArvadosModel::UnresolvableContainerError.new "docker image #{container_image.inspect} not found"
+    end
+    coll.portable_data_hash
+  end
+
   def self.find_reusable(attrs)
     candidates = Container.
-      where('command = ?', attrs[:command].to_yaml).
+      where_serialized(:command, attrs[:command]).
       where('cwd = ?', attrs[:cwd]).
-      where('environment = ?', self.deep_sort_hash(attrs[:environment]).to_yaml).
+      where_serialized(:environment, attrs[:environment]).
       where('output_path = ?', attrs[:output_path]).
-      where('container_image = ?', attrs[:container_image]).
-      where('mounts = ?', self.deep_sort_hash(attrs[:mounts]).to_yaml).
-      where('runtime_constraints = ?', self.deep_sort_hash(attrs[:runtime_constraints]).to_yaml)
-
-    # Check for Completed candidates that had consistent outputs.
-    completed = candidates.where(state: Complete).where(exit_code: 0)
-    outputs = completed.select('output').group('output').limit(2)
-    if outputs.count.count != 1
-      Rails.logger.debug("Found #{outputs.count.length} different outputs")
-    elsif Collection.
-        readable_by(current_user).
-        where(portable_data_hash: outputs.first.output).
-        count < 1
-      Rails.logger.info("Found reusable container(s) " +
-                        "but output #{outputs.first} is not readable " +
-                        "by user #{current_user.uuid}")
-    else
-      # Return the oldest eligible container whose log is still
-      # present and readable by current_user.
-      readable_pdh = Collection.
-        readable_by(current_user).
-        select('portable_data_hash')
-      completed = completed.
-        where("log in (#{readable_pdh.to_sql})").
-        order('finished_at asc').
-        limit(1)
-      if completed.first
-        return completed.first
-      else
-        Rails.logger.info("Found reusable container(s) but none with a log " +
-                          "readable by user #{current_user.uuid}")
-      end
-    end
+      where('container_image = ?', resolve_container_image(attrs[:container_image])).
+      where_serialized(:mounts, resolve_mounts(attrs[:mounts])).
+      where_serialized(:runtime_constraints, resolve_runtime_constraints(attrs[:runtime_constraints]))
+
+    # Check for Completed candidates whose output and log are both readable.
+    select_readable_pdh = Collection.
+      readable_by(current_user).
+      select(:portable_data_hash).
+      to_sql
+    usable = candidates.
+      where(state: Complete).
+      where(exit_code: 0).
+      where("log IN (#{select_readable_pdh})").
+      where("output IN (#{select_readable_pdh})").
+      order('finished_at ASC').
+      limit(1).
+      first
+    return usable if usable
 
     # Check for Running candidates and return the most likely to finish sooner.
     running = candidates.where(state: Running).
@@ -284,10 +357,12 @@ class Container < ArvadosModel
     # that a container cannot "claim" a collection that it doesn't otherwise
     # have access to just by setting the output field to the collection PDH.
     if output_changed?
-      c = Collection.
-          readable_by(current_user).
-          where(portable_data_hash: self.output).
-          first
+      c = Collection.unscoped do
+        Collection.
+            readable_by(current_user).
+            where(portable_data_hash: self.output).
+            first
+      end
       if !c
         errors.add :output, "collection must exist and be readable by current user."
       end
@@ -341,7 +416,7 @@ class Container < ArvadosModel
       act_as_system_user do
 
         if self.state == Cancelled
-          retryable_requests = ContainerRequest.where("priority > 0 and state = 'Committed' and container_count < container_count_max")
+          retryable_requests = ContainerRequest.where("container_uuid = ? and priority > 0 and state = 'Committed' and container_count < container_count_max", uuid)
         else
           retryable_requests = []
         end