I have experienced this under high capacity on my web2py sites. I have not found a solution to the issue as of yet. All I can say is I have done AB testing comparing different python web frameworks with a basic database IO, web2py just doesn't perform under high load.
Perhaps it is a coding issue with *how* you are caching your queries. When caching in ram, *each* web2py process has to cache its own version of the select. So if you reload a couple of times, apache( or your web server ) might determine it needs to spawn another process, which then in turn needs to cache all of the queries again. Solution: A) Use memcached B) or cache on disk AND ram, so if a new process starts, doesn't have it cached in ram, it will pull from the disk much quicker than re-executing the query. A is better. -- Thadeus On Wed, Jul 7, 2010 at 3:04 PM, Kuba Kucharski <kuba.kuchar...@gmail.com> wrote: > I had prepared some mini-portal, I have 1000 unique visits a day > but there is a performance issue > > bandwidth is ok > memory is ok > processor load is ok > cache.ram is set for almost all of the queries, and is set for 1 hour > > but > > sometimes, once every few hits, it loads signifacantly slower > weirdest thing is when you re-click the link it loads instantly, when > you left it working to load on itself, it is slow.. like 4 to 8 > seconds > > what could it be? where to look for an answer? I think some of you had > to see this before.. > > -- > Kuba >
