From 71ec23b55042652752e24e7ff4b7ff2f57f57355 Mon Sep 17 00:00:00 2001
From: Joseph Schorr <josephschorr@users.noreply.github.com>
Date: Wed, 18 Jan 2017 15:19:45 -0500
Subject: [PATCH] Switch QueueItem state_id to be unique after a backfill

---
 data/database.py                              |  2 +-
 ...9a_backfill_state_id_and_make_it_unique.py | 33 +++++++++++++++++++
 data/queue.py                                 |  6 +---
 3 files changed, 35 insertions(+), 6 deletions(-)
 create mode 100644 data/migrations/versions/d42c175b439a_backfill_state_id_and_make_it_unique.py

diff --git a/data/database.py b/data/database.py
index 1c84a9477..271a25c32 100644
--- a/data/database.py
+++ b/data/database.py
@@ -762,7 +762,7 @@ class QueueItem(BaseModel):
   available = BooleanField(default=True)
   processing_expires = DateTimeField(null=True)
   retries_remaining = IntegerField(default=5)
-  state_id = CharField(default=uuid_generator, index=True, unique=False)
+  state_id = CharField(default=uuid_generator, index=True, unique=True)
 
   class Meta:
     database = db
diff --git a/data/migrations/versions/d42c175b439a_backfill_state_id_and_make_it_unique.py b/data/migrations/versions/d42c175b439a_backfill_state_id_and_make_it_unique.py
new file mode 100644
index 000000000..2f65c57fd
--- /dev/null
+++ b/data/migrations/versions/d42c175b439a_backfill_state_id_and_make_it_unique.py
@@ -0,0 +1,33 @@
+"""Backfill state_id and make it unique
+
+Revision ID: d42c175b439a
+Revises: 3e8cc74a1e7b
+Create Date: 2017-01-18 15:11:01.635632
+
+"""
+
+# revision identifiers, used by Alembic.
+revision = 'd42c175b439a'
+down_revision = '3e8cc74a1e7b'
+
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy.dialects import mysql
+
+def upgrade(tables):
+    # Backfill the queueitem table's state_id field with unique values for all entries which are
+    # empty.
+    conn = op.get_bind()
+    conn.execute("update queueitem set state_id = id where state_id = ''")
+
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_index('queueitem_state_id', table_name='queueitem')
+    op.create_index('queueitem_state_id', 'queueitem', ['state_id'], unique=True)
+    # ### end Alembic commands ###
+
+
+def downgrade(tables):
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_index('queueitem_state_id', table_name='queueitem')
+    op.create_index('queueitem_state_id', 'queueitem', ['state_id'], unique=False)
+    # ### end Alembic commands ###
diff --git a/data/queue.py b/data/queue.py
index 197e45a26..af83fb806 100644
--- a/data/queue.py
+++ b/data/queue.py
@@ -263,17 +263,13 @@ class WorkQueue(object):
     # performing the update. Previously, we would check all these columns, resulting in a bunch
     # of lock contention. This change mitigates the problem significantly by only checking two
     # columns (id and state_id), both of which should be absolutely unique at all times.
-    #
-    # TODO(jschorr): Remove the extra `processing_expires` check once this has been pushed to
-    # production and every box is updating state_id.
     set_unavailable_query = (QueueItem
                              .update(available=False,
                                      processing_expires=now + timedelta(seconds=processing_time),
                                      retries_remaining=QueueItem.retries_remaining - 1,
                                      state_id=str(uuid.uuid4()))
                              .where(QueueItem.id == db_item.id,
-                                    QueueItem.state_id == db_item.state_id,
-                                    QueueItem.processing_expires == db_item.processing_expires))
+                                    QueueItem.state_id == db_item.state_id))
 
     changed = set_unavailable_query.execute()
     return changed == 1