implement smart batching

4 files changed, 94 insertions, 56 deletions

diff --git a/models/match.js b/models/match.js
index 46c4f80..ec5bdab 100644
--- a/models/match.js
+++ b/models/match.js
@@ -23,8 +23,7 @@ module.exports = function(sequelize, DataTypes) {
     },
     created_at: {
       type: DataTypes.TIME,
-      allowNull: false,
-      defaultValue: sequelize.literal('CURRENT_TIMESTAMP')
+      allowNull: false
     },
     duration: {
       type: DataTypes.INTEGER(11),
diff --git a/models/participant.js b/models/participant.js
index 9bd1ff7..9c17e74 100644
--- a/models/participant.js
+++ b/models/participant.js
@@ -59,7 +59,7 @@ module.exports = function(sequelize, DataTypes) {
     },
     gold: {
       type: DataTypes.INTEGER(11),
-      allowNull: false
+      allowNull: true
     },
     gold_mine_captures: {
       type: DataTypes.INTEGER(11),
diff --git a/models/roster.js b/models/roster.js
index 53a4411..3848c6a 100644
--- a/models/roster.js
+++ b/models/roster.js
@@ -41,10 +41,6 @@ module.exports = function(sequelize, DataTypes) {
       type: DataTypes.STRING(191),
       allowNull: false
     },
-    team_color: {
-      type: DataTypes.STRING(191),
-      allowNull: false
-    },
     turret_kills: {
       type: DataTypes.INTEGER(11),
       allowNull: false
diff --git a/worker.js b/worker.js
index 3c51624..5176189 100644
--- a/worker.js
+++ b/worker.js
@@ -6,7 +6,9 @@ var amqp = require("amqplib"),
     Seq = require("sequelize");
 
 var RABBITMQ_URI = process.env.RABBITMQ_URI || "amqp://localhost",
-    DATABASE_URI = process.env.DATABASE_URI || "sqlite:///db.sqlite";
+    DATABASE_URI = process.env.DATABASE_URI || "sqlite:///db.sqlite",
+    BATCHSIZE = process.env.PROCESSOR_BATCH || 50 * (1 + 2 + 3*2 + 3*2),
+    IDLE_TIMEOUT = process.env.PROCESSOR_IDLETIMEOUT || 500;  // ms
 
 (async () => {
     let seq = new Seq(DATABASE_URI),
@@ -14,21 +16,45 @@ var RABBITMQ_URI = process.env.RABBITMQ_URI || "amqp://localhost",
         rabbit = await amqp.connect(RABBITMQ_URI),
         ch = await rabbit.createChannel();
 
+    let queue = [],
+        timer = undefined;
+
     /* recreate for debugging
     await seq.query("SET FOREIGN_KEY_CHECKS=0");
     await seq.sync({force: true});
     */
-    //await seq.sync();
+    await seq.sync();
 
     await ch.assertQueue("process", {durable: true});
-    await ch.prefetch(1);
+    // as long as the queue is filled, msg are not ACKed
+    // server sends as long as there are less than `prefetch` unACKed
+    await ch.prefetch(BATCHSIZE);
 
     ch.consume("process", async (msg) => {
-        let match = JSON.parse(msg.content);
+        queue.push(msg);
+
+        // fill queue until batchsize or idle
+        if (timer === undefined)
+            timer = setTimeout(process, IDLE_TIMEOUT)
+        if (queue.length == BATCHSIZE)
+            process();
+    }, { noAck: false });
+
+    async function process() {
+        console.log("processing batch");
+
+        // clean up to allow processor to accept while we wait for db
+        let matchmsgs = queue.slice();
+        queue = [];
+        clearTimeout(timer);
+        timer = undefined;
 
-        // TODO commit less often if possible, avoid deadlocks
+        // BEGIN
         let transaction = await seq.transaction({ autocommit: false });
 
+        // helper to convert API response into flat JSON
+        // db structure is (almost) 1:1 the API structure
+        // so we can insert the flat API response as-is
         function flatten(obj) {
             let attrs = obj.attributes || {},
                 stats = attrs.stats || {},
@@ -42,57 +68,74 @@ var RABBITMQ_URI = process.env.RABBITMQ_URI || "amqp://localhost",
             return o;
         }
 
-        /* bring jsonapi nested response into our db structure-like shape */
-        match.rosters = match.rosters.map((roster) => {
-            roster.participants = roster.participants.map((participant) => {
-                participant.player = flatten(participant.player);
-                return flatten(participant);
-            });
-            return flatten(roster);
-        });
-        match = flatten(match);
-        console.log(match);
-
-        /* upsert everything */
-        await model.Match.upsert(match, {
-            include: [ model.Roster/*, model.Asset*/ ]
-        });
+        // UPSERT
+        try {
+            let matches = matchmsgs.map((msg) => JSON.parse(msg.content));
+            await matches.forEach(async (match) => {
+                // flatten jsonapi nested response into our db structure-like shape
+                match.rosters = match.rosters.map((roster) => {
+                    roster.participants = roster.participants.map((participant) => {
+                        participant.player = flatten(participant.player);
+                        return flatten(participant);
+                    });
+                    return flatten(roster);
+                });
+                match = flatten(match);
 
-        await match.rosters.forEach(async (roster) => {
-            roster.match_api_id = match.api_id;
-            roster.shard_id = match.shard_id;
-            await model.Roster.upsert(roster, {
-                include: [ model.Participant/*, model.Team*/ ]
-            });
+                // upsert match
+                await model.Match.upsert(match, {
+                    include: [ model.Roster, model.Asset ]
+                });
 
-            await roster.participants.forEach(async (participant) => {
-                participant.shard_id = roster.shard_id;
-                participant.player.shard_id = participant.shard_id;
-                
-                await model.Player.upsert(participant.player);
+                // upsert children
+                // before, add foreign keys and other missing information (shardId)
+                await match.rosters.forEach(async (roster) => {
+                    roster.match_api_id = match.api_id;
+                    roster.shard_id = match.shard_id;
 
+                    await model.Roster.upsert(roster, {
+                        include: [ model.Participant, model.Team ]
+                    });
 
-                participant.roster_api_id = roster.api_id;
-                participant.player_api_id = participant.player.api_id;
-                await model.Participant.upsert(participant, {
-                    include: [ model.Player ]
-                });
-            });
+                    await roster.participants.forEach(async (participant) => {
+                        participant.player.shard_id = participant.shard_id;
+                        await model.Player.upsert(participant.player);
 
-            //if (roster.team != null) model.Team.upsert(roster.team);
-        });
+                        participant.shard_id = roster.shard_id;
+                        participant.roster_api_id = roster.api_id;
+                        participant.player_api_id = participant.player.api_id;
+                        await model.Participant.upsert(participant, {
+                            include: [ model.Player ]
+                        });
+                    });
 
-        /*match.assets.forEach((asset) => {
-            model.Asset.upsert(asset);
-        });*/
+                    if (roster.team != null) {
+                        roster.team.shard_id = roster.shard_id;
+                        roster.team.roster_api_id = roster.api_id;
+                        await model.Team.upsert(roster.team);
+                    }
+                });
 
-        await transaction.commit();  // TODO rollback on err
-        ch.ack(msg);
+                await match.assets.forEach(async (asset) => {
+                    await model.Asset.upsert(asset);
+                });
+            });
 
-        await match.rosters.forEach(async (r) => {
-            await r.participants.forEach(async (p) => {
-                await ch.publish("amq.topic", p.player.name, new Buffer("process_commit"));
+            // COMMIT
+            await transaction.commit();
+            await ch.ack(matchmsgs.pop(), true);  // ack all messages until the last
+            // request child jobs, notify player
+            await matches.forEach(async (m) => {
+                await m.rosters.forEach(async (r) => {
+                    await r.participants.forEach(async (p) => {
+                        await ch.publish("amq.topic", p.player.name, new Buffer("process_commit"));
+                    });
+                });
             });
-        });
-    }, { noAck: false });
+        } catch (err) {  // TODO catch only SQL error, also catch errors in the forEach
+            console.error(err);
+            await ch.nack(matchmsgs.pop(), true, true);  // nack all messages until the last and requeue
+            // TODO don't requeue broken records
+        }
+    }
 })();