The use of "resident" instances with a low site traffic profile resulted in a different version of the pathological behavior. With periods of low usage, the scheduler decommissioned dynamic instances. When traffic picked up again, it did not use the "resident" instances. Instead, it spun up 3 or 4 dynamics causing the page load to stall out. At least, that was the case 1 month ago when I ditched the entire concept of "min idle instances" (since they appeared to always be idle and never serve traffic). I use wget to issue a single request once every 60 seconds in an attempt to have at least 1 dynamic instance always available.
I will test again to see if "idle instances" actually add any value to a site with my traffic profile. Before, they doubled or tripled the cost and did not improve the user experience at all. What is the reasoning behind not sending warmup requests? I have already starred the relevant scheduler enhancement issues. On Monday, October 22, 2012 5:51:38 AM UTC-7, Takashi Matsuo (Google) wrote: > > > Hi Carl, > > On Mon, Oct 22, 2012 at 8:26 AM, Carl Schroeder > <[email protected]<javascript:> > > wrote: > >> Has this feature been disabled? Have the requirements for it changed? >> The "Configured Services" section of my Application Settings says that >> "Warmup Requests" are enabled. >> I can find no trace of any warmup requests in my logs. >> > > Currently, the warmup requests are issued only when you have some min idle > instances settings instead of 'automatic'. > > >> >> Also, the scheduler is still acting crazy. It is starting up new java >> instances (with user facing requests) when there are dynamic instances >> alive and well and idle. 20 second response times to REST calls that >> normally return in 50ms is simply unacceptable performance. Asking my users >> to wait 40 seconds while a page that contains multiple REST calls loads is >> simply not an option. >> > > To avoid this, currently the options are > * to have sufficient number of min idle instances > or > * to make your loading requests faster > > -- Takashi > > >> >> I am back to deciding whether to port the java to python and stay on GAE, >> or trying out dynamoDB on AWS. >> >> -- >> You received this message because you are subscribed to the Google Groups >> "Google App Engine" group. >> To view this discussion on the web visit >> https://groups.google.com/d/msg/google-appengine/-/U2PyWapvPqkJ. >> To post to this group, send email to >> [email protected]<javascript:> >> . >> To unsubscribe from this group, send email to >> [email protected] <javascript:>. >> For more options, visit this group at >> http://groups.google.com/group/google-appengine?hl=en. >> > > > > -- > Takashi Matsuo | Developers Advocate | [email protected] <javascript:> > > -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To view this discussion on the web visit https://groups.google.com/d/msg/google-appengine/-/mOJNvj-ytFcJ. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.
