De-duplicate builds

When a build is triggered, it could be marked as to be skipped by builders if:

* there is already a build running/queued for the same commit

  Multiple builds of the same commit will lead to the same results. So, we skip
  it if there is one already running/queued.

* there is already a build queued for the same version

  When building a version without specifying the commit, the last commit is
  built. In this case, we can only skip the new triggered build if there is one
  build queued because both builds will just pick the same commit.

Author: humitos

readthedocs/core/utils/__init__.py

@@ -19,7 +19,7 @@
 )
 from readthedocs.doc_builder.constants import DOCKER_LIMITS
 from readthedocs.projects.constants import CELERY_LOW, CELERY_MEDIUM, CELERY_HIGH
-from readthedocs.doc_builder.exceptions import BuildMaxConcurrencyError
+from readthedocs.doc_builder.exceptions import BuildMaxConcurrencyError, DuplicatedBuildError
 
 
 log = logging.getLogger(__name__)
@@ -165,8 +165,42 @@ def prepare_build(
         # External builds should be lower priority.
         options['priority'] = CELERY_LOW
 
+    skip_build = False
+    if commit:
+        skip_build = (
+            Build.objects
+            .filter(
+                project=project,
+                version=version,
+                commit=commit,
+            ).exclude(
+                state=BUILD_STATE_FINISHED,
+            ).exists()
+        )
+    else:
+        skip_build = Build.objects.filter(
+            project=project,
+            version=version,
+            state=BUILD_STATE_TRIGGERED,
+        ).count() > 0
+    if skip_build:
+        # TODO: we could mark the old build as duplicated, however we reset our
+        # position in the queue and go back to the end of it --penalization
+        log.warning(
+            'Marking build to be skipped by builder. project=%s version=%s build=%s commit=%s',
+            project.slug,
+            version.slug,
+            build.pk,
+            commit,
+        )
+        build.error = DuplicatedBuildError.message
+        build.success = False
+        build.exit_code = 1
+        build.state = BUILD_STATE_FINISHED
+        build.save()
+
     # Start the build in X minutes and mark it as limited
-    if project.has_feature(Feature.LIMIT_CONCURRENT_BUILDS):
+    if not skip_build and project.has_feature(Feature.LIMIT_CONCURRENT_BUILDS):
         running_builds = (
             Build.objects
             .filter(project__slug=project.slug)

readthedocs/doc_builder/exceptions.py

@@ -57,6 +57,10 @@ class BuildMaxConcurrencyError(BuildEnvironmentError):
     message = ugettext_noop('Concurrency limit reached ({limit}), retrying in 5 minutes.')
 
 
+class DuplicatedBuildError(BuildEnvironmentError):
+    message = ugettext_noop('Duplicated build.')
+
+
 class BuildEnvironmentWarning(BuildEnvironmentException):
     pass
 

readthedocs/projects/tasks.py

@@ -58,6 +58,7 @@
     BuildEnvironmentWarning,
     BuildMaxConcurrencyError,
     BuildTimeoutError,
+    DuplicatedBuildError,
     MkDocsYAMLParseError,
     ProjectBuildsSkippedError,
     VersionLockedError,
@@ -542,6 +543,16 @@ def run(
             self.commit = commit
             self.config = None
 
+            if self.build['state'] == BUILD_STATE_FINISHED and self.build['error'] == DuplicatedBuildError.message:
+                log.warning(
+                    'NOOP: build is marked as duplicated. project=%s version=%s build=%s commit=%s',
+                    self.project.slug,
+                    self.version.slug,
+                    build_pk,
+                    self.commit,
+                )
+                return True
+
             if self.project.has_feature(Feature.LIMIT_CONCURRENT_BUILDS):
                 response = api_v2.build.running.get(project__slug=self.project.slug)
                 builds_running = response.get('count', 0)