Ayende @ Rahienhttp://ayende.comAyende @ RahienCopyright (C) Ayende Rahien 2004 - 2021 (c) 202660Yarden commented on Things we learned from production, part I–shutting down is hard to doRavenHD is insanely expensive.. I just can't understand anyone paying that much. If I need that capacity, I would go MySQL, jezuzhttp://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment9http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment9Tue, 06 Nov 2012 04:48:41 GMTAyende Rahien commented on Things we learned from production, part I–shutting down is hard to doJesus, We will call the IStartupTask dispose method, if it implements IDisposable, which is the hook you get when we shut down.http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment8http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment8Mon, 24 Sep 2012 12:33:58 GMTJesús commented on Things we learned from production, part I–shutting down is hard to doIs there a way for a custom bundle to participate in the shutdown process? I'm developing the update cascade bundle. This bundle might have several task in progress that probably need to be canceled.http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment7http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment7Mon, 24 Sep 2012 08:25:20 GMTAyende Rahien commented on Things we learned from production, part I–shutting down is hard to doLouis, RavenHQ is hosting a LOT of databases, and it is one of the heaviest loaded RavenDB setups. That is what I meant, it has brought to our attention a lot of interesting issues like that.http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment6http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment6Fri, 21 Sep 2012 11:30:54 GMTLouis commented on Things we learned from production, part I–shutting down is hard to doAyende, out of curiosity, could you please expand a bit on the RavenHQ Monster ;)http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment5http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment5Fri, 21 Sep 2012 11:29:31 GMTAyende Rahien commented on Things we learned from production, part I–shutting down is hard to doAndrew, I am not familiar enough with the way worker roles work. We run RavenDB in production inside IIS. Self hosted, service mode, also have some limit on shutdown, in the sense that the service manager will give an error if you take too long, but won't kill you. IIS will kill you if you take too longhttp://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment4http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment4Fri, 21 Sep 2012 11:25:59 GMTAndrew commented on Things we learned from production, part I–shutting down is hard to doIn an azure 'worker role' environment would you be able to request more time if the server is self hosted (Raven.Database.Server.HttpServer)? As an aside - would you recommend using the HttpServer instead of hosting via IIS (I wasn't sure how to use IIS hosting)http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment3http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment3Fri, 21 Sep 2012 01:03:30 GMTAyende Rahien commented on Things we learned from production, part I–shutting down is hard to doTobi, That is ASP.Net, that isn't IIS. IIS give you severe time limits.http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment2http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment2Thu, 20 Sep 2012 12:28:32 GMTtobi commented on Things we learned from production, part I–shutting down is hard to doWhen ASP.NET notifies you that it wants to unload the appdomain you can stay alive indefinitely by just not returning from your callback. Only after all callbacks have been called the countdown begins. Certainly not what the architects intended but it works.http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment1http://ayende.com/158753/things-we-learned-from-production-part-i-shutting-down-is-hard-to-do#comment1Thu, 20 Sep 2012 12:26:51 GMT