from which it seems I can monitor the backlog and then some.
For #1, I thought to write a simple tween that would write realtime stats to memcached, to be read every 5 minutes by munin for some nice graphs. I can also graph the uwsgi backlog (also there appears to be some munin plugins for this).
I'm not quite sure what to do for #2. I guess the easiest thing is start with some numbers and observe graphs for #1 and #3, change the thread per process ratio periodically and observe whether there's improvement or degradation in graphs.
> from which it seems I can monitor the backlog and then some.
> For #1, I thought to write a simple tween that would write realtime stats to memcached, to be read every 5 minutes by munin for some nice graphs. I can also graph the uwsgi backlog (also there appears to be some munin plugins for this).
> I'm not quite sure what to do for #2. I guess the easiest thing is start with some numbers and observe graphs for #1 and #3, change the thread per process ratio periodically and observe whether there's improvement or degradation in graphs.
You can use errormator wsgi client, that allows you to use free https://errormator.com service.
It will show you slow requests/slow sqlalchemy queries within your app, you would need to tweak the client to show you requests below 1s (that's min default for this in the client).
Because it's more geared towards error monitoring and finding code bottlenecks, but maybe it's a good start.
> You can use errormator wsgi client, that allows you to use free > https://errormator.com service.
> It will show you slow requests/slow sqlalchemy queries within your app,
> you would need to tweak the client to show you requests below 1s > (that's min default for this in the client).
> Because it's more geared towards error monitoring and finding code > bottlenecks, but maybe it's a good start.
Thanks for the suggestion, but the external service is not something I can use here.
> easy to use and maybe more pyramid aware. It logs
> average in sqllite db.
> view, route, url, sql min, max average
> it could be useful to monitor projects
> Hope to be helpfull
Thanks, this seems to be what I need. I originally meant to use memcached for performance reasons. I need this for constant monitoring not just to troubleshoot, so I'm afraid sqlite might become a bottleneck, but we'll see.
you're right, It's just a starter kit to help young developers
getting started with pyramid, and for the
moment I only target alchemy scaffold
(get inspired from dozer project and so one ...).
my goal :
- a project with tiny finger print (low package dependency)
- scaffold agnostic, it should work with url dispatch or traversal
o disable sql reports for traversal
- be easy to configure and high level information
- high performance, reduce the latency induced.
(I've hardcoded sql to improve that, but it's a mess => brown bag)
- offer some kind of api to work with different backend ...
I hope to get some times in few week to push another release
Feel free to fork the project and propose a work around, I'll be
happy to go further.
>> easy to use and maybe more pyramid aware. It logs
>> average in sqllite db.
>> view, route, url, sql min, max average
>> it could be useful to monitor projects
>> Hope to be helpfull
> Thanks, this seems to be what I need. I originally meant to use > memcached for performance reasons. I need this for constant monitoring > not just to troubleshoot, so I'm afraid sqlite might become a > bottleneck, but we'll see.
