Web.cacheModule and Expiring DB connections

The Story

A while ago I posted about neko.web.cacheModule, a way to make your site run dramatically faster on mod_neko or mod_tora. After spending some time making ufront compatible with this mode of running, I was excited to deploy my app to our schools with the increased performance.

I deployed, ran some tests, seemed happy, went home and slept well. And woke up the next morning to a bunch of 501 errors.

The Error

The error message was coming from the MySQL connection: “Failed to send packet”.

At first I just disabled caching, got everyone back online, and then went about trying to isolate the issue. It took a while before I could reproduce it and pinpoint it. I figured it was to do with my DB connection staying open between requests, thanks to the new module caching.

A google search showed only one Haxe related result – someone on IRC that mentioned when they sent too many SQL queries it sometimes bombed out with this error message. Perhaps leaving it open eventually overran some buffer and caused it to stop working? Turns out this was not the case, I used `ab` (Apache Benchmark tool) to query a page 100,000 times and still didn’t see the error.

Eventually I realised it was to do with the MySQL Server dropping the connection after a period of inactivity. The `wait_timeout` variable was set to 28800 by default: so after 8 hours of inactivity.  So long enough that I didn’t notice it the night before, but short enough that the timeout occured overnight while all the staff were at home or asleep… So the MySQL server dropped the connection, and my Haxe code did not know to reconnect it. Whoops.

The Solution

I looked at Nicolas’s hxwiki source code, which runs the current Haxe site for some inspiration on the proper way to approach this for `neko.Web.cacheModule`. His solution: use http://api.haxe.org/sys/db/Transaction.html#main. By the looks of it, this will wrap your entire request in an SQL transaction, and should an error be thrown, it will rollback the transaction. Beyond that, it will close the DB connection after each request. So we have a unique DB connection for each request, and close it as soon as the request is done.

My source code looks like this:

class Server
{
  static var ufApp:UfrontApplication;

  static function main() {
    #if (neko && !debug) neko.Web.cacheModule(main); #end
    
    // Wrap all my execution inside a transaction
    sys.db.Transaction.main( Mysql.connect(Config.db), function() {
      init();
      ufApp.execute();
    });
  }

  static function init() {
    // If cacheModule is working, this will only run once
    if (ufApp==null) {
      UFAdminController.addModule( "db", "Database", new DBAdminController() );
      ufApp = new UfrontApplication({
        dispatchConfig: Dispatch.make( new Routes() ),
        remotingContext: Api,
        urlRewrite: true,
        logFile: "log/ufront.log"
      });
    }
  }
}

You cannot use @:build inside a macro : make sure that your enum is not used in macro

If you ever come across this error, it can be a bit cryptic to understand.

Here’s some sample code that produces the error:

import haxe.macro.Expr;
import haxe.macro.Context;
import neko.Lib;

class BuildInMacroTests 
{
    public static function main() {
        var s:MyModel = new MyModel();
        MyMacro.doSomething( s );
    }
}

class MyModel extends sys.db.Object // has a build macro
{
    var id:Int;
    var name:String;
}

private class MyMacro 
{
    public static macro function doSomething(output : Expr) : Expr {
        return output;
    }
}

I’m still not sure I have a full grasp on it, but here’s a few pointers:

  • It’s not necessarily an enum – it can happen with a class or interface that has a “@:build” macro also.
  • The basic sequence happens in this order:
    • BuildInMacrosTest.main() is the entry point, so the compiler starts there and starts processing.
    • When it hits “var s:MyModel”, it does a lookup of that type, realizes the class needs a build macro to run, and runs the build macro.
    • When it hits “MyMacro.doSomething()”, it types the expression, and realizes that it is a macro call.  To run the macro, it must find the macro class, load it into the macro context, and execute it.
    • It finds the macro class, it’s in this file.
    • It tries to load this module (hx file) into the macro context, so it goes through the whole process of typing it again.
    • As it’s loading it into the macro context, it hits the “MyModel” build macro again, which it can’t do at macro time, so it spews out the error.
  • The basic solutions:
    • Wrap your build macro declarations in conditionals:
      #if !macro @:build(MyMacro.build()) #end class Object { ... }
    • Wrap anything that is not a macro in conditionals:
      #if !macro 
        class BuildInMacroTests {}
        class MyModel {}
      #else
        class MyMacro {}
      #end
    • Keep your macros in seperate files:
      BuildInMacroTests.hx:
          class BuildInMacroTests {
              public static function main() {
                  var s:MyModel = new MyModel();
                  MyMacro.doSomething( s );
              }
          }
      
          class MyModel extends sys.db.Object {
              var id:Int;
              var name:String;
          }
      
      MyMacro.hx:
          import haxe.macro.Expr;
          import haxe.macro.Context;
          class MyMacro {
              public static macro function doSomething(output : Expr) : Expr {
                  return output;
              }
          }

A combination of the 1st and 3rd solutions is probably usually the cleanest.

Good luck!

A Haxe/JS Debugging Tip

When targetting Haxe JS in Haxe 3, the output is “modern” style, which means, to prevent polluting the global namespace or conflicting with other libraries, the output is wrapped in a simple function:

(function() {})()

Which is great.  And if you place breakpoints in your code, using:

js.Lib.debug();

Then your browser will launch the debugger inside this Haxe context, and you have access to all your classes etc.  But what if you want to fire up the browser’s JS console, and gain arbitrary access to something in your Haxe code?  Because it’s all hidden inside that function, you can’t.

Unless you have a workaround, which looks like this:

class Client 
{
    @:expose("haxedebug") @:keep
    public static function enterDebug()
    {
        js.Lib.debug();
    }
}

What’s going on: we have a class, and a function: “enterDebug”.  This function can go in any class that you use, really – it doesn’t have to be in Client or your Main class or anything.

The “js.Lib.debug()” statement launches the debugger in the haxe context, as described before.  But the “@:expose” metadata exposes this function outside of the Haxe context.  The string defines what we expose it as: rather than the default “Client.enterDebug()”, we’ll just have “haxedebug()”.  And the “@:keep” metadata makes sure this won’t get deleted by the compilers Dead Code Elimination, even though we may never explicitly call this function in our code.

Now that we’ve done that, recompile, and voilà!  You can type “haxedebug()” into the Javascript console and start fiddling around inside the Haxe context.  Enjoy.