Long time ago I wrote some kind of profiler for T5, its not JMX, though it also stores results in-memory and there was a T5 page with profiling results. The idea was to advise each method invocation and measure statistics, like instrumenting profilers do.
https://github.com/dmitrygusev/ping-service/blob/master/ping-service/src/dmitrygusev/ping/services/AppModule.java#L430 https://github.com/dmitrygusev/ping-service/blob/master/ping-service/war/Profiler.tml https://github.com/dmitrygusev/ping-service/blob/master/ping-service/src/dmitrygusev/ping/pages/Profiler.java https://github.com/dmitrygusev/ping-service/tree/master/ping-service/src/dmitrygusev/ping/services/profiler On Thu, Jan 19, 2012 at 11:32, Kalle Korhonen <kalle.o.korho...@gmail.com>wrote: > Thanks Josh, sounds like we have a winner since there are no other > contestants. I remember seeing your announcement at the time, but my > quick search didn't even turn it up now. I'll check it out. > > Kalle > > > On Tue, Jan 17, 2012 at 10:21 AM, Josh Canfield <joshcanfi...@gmail.com> > wrote: > > Hey Kalle. > > > > I started tapestry-monitoring so that I could keep track of specific > > methods runtime (loading data from the database etc). > > > > https://github.com/joshcanfield/tapestry-monitoring > > > > I've had visions of extending it but I haven't needed other features > > enough to bump up the priority. > > > > I think it'd be pretty straight forward to add page level metrics. > > I've also considered adding a companion module that reads the > > monitoring information and presents a UI, but the production > > environment for my current project uses a third party app to read the > > JMX values and track them. > > > > Josh > > > > On Mon, Jan 16, 2012 at 6:58 PM, Kalle Korhonen > > <kalle.o.korho...@gmail.com> wrote: > >> Since Java is so stupendously fast (really!) I've been more and more > >> focused on scaling up lately, as opposed to the "easy cop-out" by > >> scaling out, in an effort to reduce running costs. If any of you have > >> tried running a bigger site on GAE, you know what I mean - you think > >> it's cheap in the beginning, but pretty soon you need reserved > >> instances, more memory, increased bandwidth limits etc. and it all > >> adds up pretty quick. Now there's a lot different tools for testing > >> website throughput and other performance aspects, but I'm looking for > >> something Tapestry-specific so that serving static resources and other > >> irrelevant activities wouldn't skew my numbers. I already have > >> something home-grown, but it'd be great if there was a comprehensive > >> yet lightweight package for gathering long-term data and performance > >> averages for serving page requests, ideally attaching itself to the > >> request pipeline. I'm trying to push page handling averages including > >> database access down to 10ms or below for at least 1 million requests > >> a day per system. So I need comprehensive long-term data to know where > >> the hotspots are. If somebody has something half-baked written > >> already, I wouldn't mind participating in taking it to a solid, > >> generally useful module for all Tapestry users. > >> > >> Kalle > >> > >> --------------------------------------------------------------------- > >> 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 > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: users-unsubscr...@tapestry.apache.org > For additional commands, e-mail: users-h...@tapestry.apache.org > > -- Dmitry Gusev AnjLab Team http://anjlab.com
