My method to cache templates at the warmup handler: calling .get_template 
method of the jinja2 environment
It does work, however it's not recursive as I thought, so you need to 
manually call .get_template for each sub template / templates that get 
included/imported

I've extended the templates that I ping at warmup handler, this bumped the 
warmup handler time from 2 seconds to 4+/5 seconds
I was missing 2 templates that are used at the request that takes 26 
seconds to render the initial template (0 seconds for the later calls, 
since it's cached) - the initial one reduced it to 12 seconds, and the 
other one to 0 seconds

*What I've learned*
*1) *Templates take toooooo long to render with F2 instances, extremely 
long, like ~ 1 seconds for a not too complicated template
*2)* Templates take *5X-10X* time to render/get cached on normal requests 
compared to the warmup handlers
*3)* There might be something wrong with how the first requests are 
replied, extending the delay to the neighboring requests (my experiments 
didn't prove this)

Things were pretty consistent in my trials, the request times were always 
consistent and I've eliminated the delay issue by pre-caching the jinja2 
templates

What do you guys think? Isn't 1 second a lot of time to render a template? 
(>>1 seconds if you do it during multi-threading)

-- 
You received this message because you are subscribed to the Google Groups 
"Google App Engine" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/google-appengine.
For more options, visit https://groups.google.com/d/optout.

Reply via email to