fix(serverless-api/logs): keeps the most recent logs in log cache

Another note, I discovered while testing this that _read() is called multiple times and lead to extra API calls via _poll(). This was because when calling _read() with an interval set, the code would make a one off call. Now, if an interval is set already, then no extra polls will be made.

Author: philnash

packages/serverless-api/src/streams/logs.ts

@@ -9,6 +9,7 @@ export class LogsStream extends Readable {
   private _pollingCacheSize: number;
   private _interval: NodeJS.Timeout | undefined;
   private _viewedSids: Set<Sid>;
+  private _viewedLogs: Array<{ sid: Sid; dateCreated: Date }>;
 
   constructor(
     private environmentSid: Sid,
@@ -19,6 +20,7 @@ export class LogsStream extends Readable {
     super({ objectMode: true });
     this._interval = undefined;
     this._viewedSids = new Set();
+    this._viewedLogs = [];
     this._pollingFrequency = config.pollingFrequency || 1000;
     this._pollingCacheSize = config.logCacheSize || 1000;
   }
@@ -54,13 +56,35 @@ export class LogsStream extends Readable {
       // The logs endpoint is not reliably returning logs in the same order
       // Therefore we need to keep a set of all previously seen log entries
       // In order to avoid memory leaks we cap the total size of logs at 1000
-      // If the new set is larger we'll instead only use the SIDs from the current
-      // request.
-      if (logs.length + this._viewedSids.size <= this._pollingCacheSize) {
-        logs.map(log => log.sid).forEach(sid => this._viewedSids.add(sid));
-      } else {
-        this._viewedSids = new Set(logs.map(log => log.sid));
-      }
+      // (or the set pollingCacheSize).
+      //
+      // We store an array of the logs' SIDs and created dates.
+      // Then when a new page of logs is added, we find the unique logs, sort by
+      // date created, newest to oldest, and chop off the end of the array (the
+      // oldest logs) leaving the most recent logs in memory. We then turn that
+      // into a set of SIDs to check against next time.
+
+      // Creates a unique set of log sids and date created from previous logs
+      // and new logs by stringifying the sid and the date together.
+      const viewedLogsSet = new Set([
+        ...this._viewedLogs.map(
+          log => `${log.sid}-${log.dateCreated.toISOString()}`
+        ),
+        ...logs.map(log => `${log.sid}-${log.date_created}`),
+      ]);
+      // Then we take that set, map over the logs and split them up into sid and
+      // date again, sort them most to least recent and chop off the oldest if
+      // they are beyond the polling cache size.
+      this._viewedLogs = [...viewedLogsSet]
+        .map(logString => {
+          const [sid, dateCreated] = logString.split('-');
+          return { sid, dateCreated: new Date(dateCreated) };
+        })
+        .sort((a, b) => b.dateCreated.valueOf() - a.dateCreated.valueOf())
+        .slice(0, this._pollingCacheSize);
+      // Finally we create a set of just SIDs to compare against.
+      this._viewedSids = new Set(this._viewedLogs.map(log => log.sid));
+
       if (!this.config.tail) {
         this.push(null);
       }
@@ -70,10 +94,12 @@ export class LogsStream extends Readable {
   }
 
   _read() {
-    if (this.config.tail && !this._interval) {
-      this._interval = setInterval(() => {
-        this._poll();
-      }, this._pollingFrequency);
+    if (this.config.tail) {
+      if (!this._interval) {
+        this._interval = setInterval(() => {
+          this._poll();
+        }, this._pollingFrequency);
+      }
     } else {
       this._poll();
     }