Hi Alex & Co.,I've found that even running off appengine's own cron facility (every 2 minutes), there is the occasional time, looks like once every 3 hours, that the app is unloaded and the 8 - 10 second response delay occurs.
Here is a portion of the logs. You'll see at 01:23PM the app loads fine but even 2 minutes later it takes nearly 10 seconds. Looks like the the paid reserve instance when it becomes available will be the most reliable way to keep the app "warm".
I'm only fetching / Chris• 07-05 01:25PM 06.806 / 200 6096ms 9508cpu_ms 2kb AppEngine-Google; (+http://code.google.com/appengine) See details 0.1.0.1 - - [05/Jul/2010:13:25:12 -0700] "GET / HTTP/1.1" 200 2614 - "AppEngine-Google; (+http://code.google.com/appengine)" "tapgae1.appspot.com" ms=6097 cpu_ms=9508 api_cpu_ms=0 cpm_usd=0.264438 queue_name=__cron task_name=70250254b79e5ded197224d2233e2917 loading_request=1
/snipped tapestry logging of a reload •• 07-05 01:23PM 06.757 / 200 30ms 38cpu_ms 2kb AppEngine-Google; (+http://code.google.com/appengine) See details 0.1.0.1 - - [05/Jul/2010:13:23:06 -0700] "GET / HTTP/1.1" 200 2614 - "AppEngine-Google; (+http://code.google.com/appengine)" "tapgae1.appspot.com" ms=30 cpu_ms=38 api_cpu_ms=0 cpm_usd=0.001383 queue_name=__cron task_name=70250254b79e5ded197224d2233e2917
• I 07-05 01:23PM 06.782
[tapgae1/3.343092403328231167].<stdout>: [INFO] AppModule.TimingFilter
Request time: 10 ms
On 06/07/2010, at 12:17 PM, Alex Kotchnev wrote:
Thiago, you're probably right, loading a single page might be enough . Imentioned what I do and it seems to work OK - I didn't necessarily implythat it was the only or the best way of keeping the app warm.On the benefits of keeping GAE from shutting down the app server : I've noticed that if I don't keep it warm, I easily spend the same 20% of the quota just starting and stopping the app , with the downside that the firstpage load takes forever (e.g. 30 secs). Regards, Alex K On Sun, Jul 4, 2010 at 6:31 PM, Thiago H. de Paula Figueiredo < thiag...@gmail.com> wrote:On Sun, 04 Jul 2010 15:31:32 -0300, Howard Lewis Ship <hls...@gmail.com >wrote: It seems to me that just getting the root index page would accomplishmost of the goal: keeping the whole application from being unloaded. Your spider will do that, plus keep the page pool filled with pages. Did you experiment with just getting the root index page?Wouldn't any page work as well as the root page to keep GAE from deallocation the application? I would create a page that returns nothing for that. -- Thiago H. de Paula FigueiredoIndependent Java, Apache Tapestry 5 and Hibernate consultant, developer,and instructor Owner, Ars Machina Tecnologia da Informação Ltda. http://www.arsmachina.com.br --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tapestry.apache.org For additional commands, e-mail: users-h...@tapestry.apache.org
--------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tapestry.apache.org For additional commands, e-mail: users-h...@tapestry.apache.org
