Switch the build logs archiver to a more performant query

Fixes #459

data/model/build.py

@@ -4,7 +4,7 @@ from peewee import JOIN_LEFT_OUTER
 from datetime import timedelta, datetime
 
 from data.database import (BuildTriggerService, RepositoryBuildTrigger, Repository, Namespace, User,
-                           RepositoryBuild, BUILD_PHASE, db_for_update)
+                           RepositoryBuild, BUILD_PHASE, db_for_update, db_random_func)
 from data.model import (InvalidBuildTriggerException, InvalidRepositoryBuildException,
                         db_transaction, user as user_model)
 
@@ -163,11 +163,23 @@ def cancel_repository_build(build, work_queue):
     return True
 
 
-def archivable_buildlogs_query():
+def get_archivable_build():
   presumed_dead_date = datetime.utcnow() - PRESUMED_DEAD_BUILD_AGE
-  return (RepositoryBuild
-          .select()
+  candidates = (RepositoryBuild
+          .select(RepositoryBuild.id)
           .where((RepositoryBuild.phase == BUILD_PHASE.COMPLETE) |
                  (RepositoryBuild.phase == BUILD_PHASE.ERROR) |
                  (RepositoryBuild.started < presumed_dead_date),
-                 RepositoryBuild.logs_archived == False))
+                 RepositoryBuild.logs_archived == False)
+          .limit(50)
+          .alias('candidates'))
+
+  try:
+    return (RepositoryBuild
+             .select(candidates.c.id)
+             .from_(candidates)
+             .order_by(db_random_func())
+             .get())
+  except RepositoryBuild.DoesNotExist:
+    return None
+