If this isn't an example of mis-configuration (somehow having pushed a "1/m" queue definition in the past), then it probably warrants a post to the Public Issue Tracker <http://code.google.com/p/google-appengine/issues/list>. The information you might include would be the code behind /tasks/EventProcessor, along with the app id of the affected app and the queue name, along with other potentially-related information, such as the QPS on the instances which are handling the requests.
Posting there will quickly lead to a reply from someone in Community Support doing triage of Public Issue Tracker issues. Whether this has always been this way, or started recently, would also be some useful information. It might have been hinted by your post, but I can't determine for sure. Best of luck! On Wednesday, December 16, 2015 at 1:06:21 PM UTC-5, Thanasis Delenikas wrote: > > Hello, > > I have a Java application which happily runs in the free tier of Google > App Engine. In the past I've encountered some spikes - from that moment > I've enabled the paid tier with a small daily budget. So, although I have > allocated some $$$ for daily spikes, my application runs in the free-quote > range for most of its life. > > My problem is task queues. > > For example, I have a queue with bucket size "20", rate "1/s" and max > concurrency "1". > > When this queue gets something like 10-20 tasks (each task takes some > seconds to execute), I see an actual processing rate of "1 / minute" - yes, > per minute. In the same time, the running instance does no work at all and > GAE makes no attempt to start a second instance - to "maybe" boost the > processing rate in order to try and align with the "1/s" configured rate. > > Here is a log of the workload and the processing ratio - check the hours / > minutes: > > I 2015-12-16 11:55:50.488 200 0 B 3.19 s > /tasks/EventProcessor?eventId=W475967 > I 2015-12-16 11:54:06.255 200 0 B 14.67 s > /tasks/EventProcessor?eventId=W476837 > I 2015-12-16 11:53:10.281 200 0 B 4.65 s > /tasks/EventProcessor?eventId=W476829 > I 2015-12-16 11:52:11.583 200 0 B 9.21 s > /tasks/EventProcessor?eventId=W476800 > I 2015-12-16 11:51:11.480 200 0 B 5.25 s > /tasks/EventProcessor?eventId=W476697 > I 2015-12-16 11:50:11.377 200 0 B 5.47 s > /tasks/EventProcessor?eventId=W476678 > I 2015-12-16 11:49:11.311 200 0 B 4.68 s > /tasks/EventProcessor?eventId=W476660 > I 2015-12-16 11:48:11.216 200 0 B 10.97 s > /tasks/EventProcessor?eventId=W476523 > I 2015-12-16 11:47:11.143 200 0 B 2.76 s > /tasks/EventProcessor?eventId=W476522 > I 2015-12-16 11:46:11.048 200 0 B 6.3 s > /tasks/EventProcessor?eventId=W476394 > I 2015-12-16 11:45:10.945 200 0 B 3.41 s > /tasks/EventProcessor?eventId=W476391 > I 2015-12-16 11:44:10.921 200 0 B 6.2 s > /tasks/EventProcessor?eventId=W476361 > I 2015-12-16 11:43:10.817 200 0 B 2.68 s > /tasks/EventProcessor?eventId=W476359 > I 2015-12-16 11:42:10.713 200 0 B 6.21 s > /tasks/EventProcessor?eventId=W476345 > I 2015-12-16 11:41:10.696 200 0 B 438 ms > /tasks/EventProcessor?eventId=W476252 > I 2015-12-16 11:40:10.593 200 0 B 6.41 s > /tasks/EventProcessor?eventId=W476130 > F 2015-12-16 11:40:01.720 500 0 B 8.8 s > /tasks/EventProcessor?eventId=W475967 > > > Why doesn't GAE respect the defined ratio? Am I doing something very > wrong? Have I missed anything? > Does this happen because the actual processing takes seconds instead of ms? > > Should I leave automatic scaling behind and move to custom scaling? > > Thanks! > > -- 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 https://groups.google.com/group/google-appengine. To view this discussion on the web visit https://groups.google.com/d/msgid/google-appengine/fa031d62-2136-4fb7-8d76-4ec221ad3f0f%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
