Hi Massimo --
Yes, I added the response.toolbar() to my ajax response and placed it at
the bottom of the served page. It basically confirmed the Postgres tools
-- my page is served in around 30ms for a typical request (actually faster
than "explain analyze" predicted). But the wall clock says it takes
between 2-10 seconds to serve the page. Maybe web2py is so fast I'm
experiencing time dilation. ;-)
My guess is that the slowness comes from the interplay of the pieces in my
total setup. So far I have:
1. A 6-core AMD processor running Linux
2. Postgres running on top of it
3. Virtualbox running, with 3 guests.
4. Web2py and nginx in a virtualbox guest, on Linux
5. "Host-only" networking for speedy web2py-postgress communications
between VB host and guest
6. "Bridged" networking for external communication to web2py instance
So far the loading is one user, so I know I'm not getting too many
requests. The networking setup works very well since i use the same
channels for development on the web2py virtualbox -- host-only when at my
Linux box and bridged when at a different computer. Response is snappy and
immediate in all cases. Only the web2py-nginx-postgres path seems to be
excessively slow, and only when I'm using jQuery AJAX requests.
On Tuesday, July 2, 2013 4:49:31 AM UTC-7, Massimo Di Pierro wrote:
>
> Did you try the {{=response.toolbar()}} ?
>
> On Tuesday, 2 July 2013 05:18:27 UTC-5, Joe Barnhart wrote:
>>
>> I have an issue, but my question is really a "meta issue" about the
>> issue...
>>
>> I'm developing a large database application which uses a postgres server
>> which is separate from the web2py installation (on nginx). When geting
>> pages currently the time to fetch a page is 2-10 seconds! I have profiled
>> the database -- it's returning the data in about 100ms. I profiled the
>> controller (including the database) and it's responding in 200-400ms. So
>> my task is to find the extra 1.5 to 9.5 seconds.
>>
>> Which leads to my question -- how to debug issues like this? I'm
>> familiar with postgres and the tools there to analyze and explain a query.
>> I can instrument my web2py code and have it tell me the resulting time to
>> run a controller. But the overall application, with the interaction of two
>> computers, browsers, etc. is just too fragmented for me to see where the
>> time is going and it's too complicated to post a simple example here and
>> have one of you geniuses tell me the problem.
>>
>> I really need some strategies for debugging these system issues myself.
>> Any tips or tools I should be looking at? (For example I have an use
>> WingIDE which has been very helpful with some issues but not this one so
>> much.)
>>
>> Joe
>>
>
--
---
You received this message because you are subscribed to the Google Groups
"web2py-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
