I am trying to reason with figures and percentage in my daily quotas. Not much of progress.
I'm using GAE only for api for my android app, for saving user's items count, hence no static files or so.
Just small packets coming back and forth between them.
No crons, taskqueue, or any fancy functions at all is used. Just endpoints apis and datastore are.
So far, up until yesterday, per day, the whole requests were about 500, more or less, 18 instance hours, one-digit percentage on read/write operations.
All of a sudden, today in the morning (fyi, from east Asia), requests surged above 50,000, instance hours above 83, and ... please check attached links for snapshots below.
For instance hours, I guess it is possible, since I am using autoscaled option for them. Maybe GAE brought up a higher class of instance for better throughput. Just guessing.
But, I cannot even imagine how on earth requests became 50,000 over a night. Over 100 times! At first sight I thought I hit the jackpot, but logs tell differently. I, manually, counted how many logs for requests were came in on GAE-Logs during the past 24 hours. The count of requests were merely over 1000. (it's about twice as much more compared to yesterday's) It seems something else just drained my quotas, for something else I don't know... or for nothing...?
Also, 3 million ops of Datastore Read Operations for today makes no sense at all to me. It's three times amount of freely-allowed quotas for read operations. Until yesterday, it showed around 3% or 4%.
I am very confused. Would you please check the images links below and share your thoughts with me?
Thank you so much!
--------------------------------------updated----------------------------------------
I checked if my free quotas got reset after 6 hours for another day. But it starts where it was, like the capture below...Checked if there were massive requests from users or something. No luck. Any possibility that it might be a bug, unless there is a possible explanation?
I'm answering to my question:
I contacted Google Cloud Platform Support. It turned out that the dev. team could not figure out the real cause because they did not keep the logs. They presumed it is very likely to be UI bug.