From a159bd3e7746af1dc7e8a381456aee729cda5f45 Mon Sep 17 00:00:00 2001
From: Jake Moshenko <jake.moshenko@coreos.com>
Date: Mon, 24 Apr 2017 13:37:25 -0400
Subject: [PATCH] Resolve race condition between multiple log archivers

---
 data/model/build.py          | 10 ++++++++++
 workers/buildlogsarchiver.py | 30 ++++++++++++++++--------------
 2 files changed, 26 insertions(+), 14 deletions(-)

diff --git a/data/model/build.py b/data/model/build.py
index 346d9ebef..8d2dc6048 100644
--- a/data/model/build.py
+++ b/data/model/build.py
@@ -248,3 +248,13 @@ def get_archivable_build():
     return RepositoryBuild.get(id=found_id)
   except RepositoryBuild.DoesNotExist:
     return None
+
+
+def mark_build_archived(build_uuid):
+  """ Mark a build as archived, and return True if we were the ones who actually
+      updated the row. """
+  return (RepositoryBuild
+          .update(logs_archived=True)
+          .where(RepositoryBuild.uuid == build_uuid,
+                  RepositoryBuild.logs_archived == False)
+          .execute()) > 0
diff --git a/workers/buildlogsarchiver.py b/workers/buildlogsarchiver.py
index 0c17c7646..7b9f2ba98 100644
--- a/workers/buildlogsarchiver.py
+++ b/workers/buildlogsarchiver.py
@@ -15,6 +15,7 @@ MEMORY_TEMPFILE_SIZE = 64 * 1024  # Large enough to handle approximately 99% of
 
 logger = logging.getLogger(__name__)
 
+
 class ArchiveBuildLogsWorker(Worker):
   def __init__(self):
     super(ArchiveBuildLogsWorker, self).__init__()
@@ -38,22 +39,23 @@ class ArchiveBuildLogsWorker(Worker):
       'logs': entries,
     }
 
-    with CloseForLongOperation(app.config):
-      with SpooledTemporaryFile(MEMORY_TEMPFILE_SIZE) as tempfile:
-        with GzipFile('testarchive', fileobj=tempfile) as zipstream:
-          for chunk in StreamingJSONEncoder().iterencode(to_encode):
-            zipstream.write(chunk)
+    if length > 0:
+      with CloseForLongOperation(app.config):
+        with SpooledTemporaryFile(MEMORY_TEMPFILE_SIZE) as tempfile:
+          with GzipFile('testarchive', fileobj=tempfile) as zipstream:
+            for chunk in StreamingJSONEncoder().iterencode(to_encode):
+              zipstream.write(chunk)
 
-        tempfile.seek(0)
-        log_archive.store_file(tempfile, JSON_MIMETYPE, content_encoding='gzip',
-                               file_id=to_archive.uuid)
+          tempfile.seek(0)
+          log_archive.store_file(tempfile, JSON_MIMETYPE, content_encoding='gzip',
+                                 file_id=to_archive.uuid)
 
-    to_update = model.build.get_repository_build(to_archive.uuid)
-    to_update.logs_archived = True
-    to_update.save()
-
-    build_logs.expire_status(to_update.uuid)
-    build_logs.delete_log_entries(to_update.uuid)
+    we_updated = model.build.mark_build_archived(to_archive.uuid)
+    if we_updated:
+      build_logs.expire_status(to_archive.uuid)
+      build_logs.delete_log_entries(to_archive.uuid)
+    else:
+      logger.debug('Another worker pre-empted us when archiving: %s', to_archive.uuid)
 
 
 if __name__ == "__main__":