I am not sure that they are not using tasks or crons for auto-scaling threshold. Simply because because average latency is simply arithmetic mean of every instance latency. I have simple user-facing request that starts chain of 15 tasks. I have cold start for this user-facing request and second cold start for fifth task. With two instances now servicing only tasks, average latency is changed because of every task instance latency and although my task usually have let's say 200 ms latency, because of this pending_ms time average latency is 662.2 ms. But who can predict what will happen with 50 clients (this is web application, not web page) how can I be sure that in this situation there will not be too big pending time, maybe even DeadlineExceededException only because of this pending_ms time included in average latency.
On Oct 26, 3:27 am, Julian Namaro <namarojul...@gmail.com> wrote: > The first number is used for average latency, not cpu_ms, and the auto- > scaling threshold is computed only from user-facing request, not tasks > or crons. > > Your numbers seems quite good, nothing to worry about I think. Are you > experiencing a specific problem ? > > On Oct 23, 12:22 am, Matija <matija.jerko...@gmail.com> wrote: > > > > > 3. If my request has these performance data: 188ms 5269cpu_ms > > 5133api_cpu_ms. > > > Does 188 ms counts for average latency (for new instances) or 5269 ms > > (cpu time) ? -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To post to this group, send email to google-appeng...@googlegroups.com. To unsubscribe from this group, send email to google-appengine+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.