Request was aborted errors

108 views
Skip to first unread message

Mind

unread,
Jul 5, 2012, 10:45:21 PM7/5/12
to google-a...@googlegroups.com
I have a GAE app written in Python that provides the back end for a multi-player Android game.  The game just got 'featured' on Google Play.

Usage has significantly increased, and I am starting to see errors like the following:

  1. 68.48.166.111 - - [05/Jul/2012:19:33:27 -0700] "GET /client/gs HTTP/1.1" 500 0 - "Apache-HttpClient/UNAVAILABLE (java 1.4)" "sven-wordhero.appspot.com" ms=21604 cpu_ms=0 api_cpu_ms=0 cpm_usd=0.000031 pending_ms=10963 
  2. I 2012-07-05 22:33:27.594
    Request was aborted after waiting too long to attempt to service your request.
I don't have quota problems (using 1/10th of daily quota atm) and I don't have an upper limit on the number of instances.  The code is Python 2.7 and marked thread-safe.

Here is the code that is failing to respond (in about 8% of cases)... actually, I don't even think this code gets a chance to run (GAE just fails to respond to the HTTP GET?):

        lg_ref = LastGame.get_by_key_name('current_game')
        results = lg_ref.gameStorageKey
           
        out = {'u': results.players,
               's': results.scores,
              }
        return self.response.out.write(json.dumps(out))

Since the game is synchronized, I have about 500 users hit this URL within a 5 second window.  No, I cannot use memcache.  It is BROKEN.  It randomly fails to update the results (even with expiry time set!) and then provides the results from the PREVIOUS game.  Users don't like that!

Takashi Matsuo

unread,
Jul 6, 2012, 1:43:20 AM7/6/12
to google-a...@googlegroups.com

Hi Mind,

Generally, I would suggest settings appropriate number of 'Min Idle Instances' for such errors.
However, it's not possible to give a proper advise from information you've provided.

What is your app-id?
How long is the average latency of your program?
Have you used appstats for profiling?

Re: memcache
Don't rely entirely on the memcache service, because currently the memcache contents might be evicted any time.

-- Takashi



--
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/-/newhM5g7zPMJ.
To post to this group, send email to google-a...@googlegroups.com.
To unsubscribe from this group, send email to google-appengi...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.



--
Takashi Matsuo

Richard

unread,
Jul 6, 2012, 9:50:10 PM7/6/12
to google-a...@googlegroups.com
Hi Takashi,

App id is sven-wordhero.  Max latency is set to 500msec.

I am seeing errors in the dashboard atm with 30 instances running.  The error rate is 3% at the moment.  This is with only 500 requests.  I actually cache the data locally in a global var (along with a datetime) to try reduce the number of database reads to help with this problem..... but the problem is still there!

I have not used app-stats in a long time.

This is a simple 'get', json up the results and return them.  I really don't think there should be a problem here!

-Richard
To post to this group, send email to google-appengine@googlegroups.com.
To unsubscribe from this group, send email to google-appengine+unsubscribe@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.



--
Takashi Matsuo
To post to this group, send email to google-appengine@googlegroups.com.
To unsubscribe from this group, send email to google-appengine+unsubscribe@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.



--
Takashi Matsuo
To post to this group, send email to google-appengine@googlegroups.com.
To unsubscribe from this group, send email to google-appengine+unsubscribe@googlegroups.com.

For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.



--
Takashi Matsuo

Andrew Mackenzie

unread,
Sep 25, 2012, 7:04:18 AM9/25/12
to google-a...@googlegroups.com
This morning (September 25, 2012) I was performing testing on an appID. 

Until about 12.30PM, CET there was no sign of errors.

Since then (now 1PM) I get this error reported about 30%-50% of the time on my simple GET request to my simple test servlet that does nothing, except sleep for a period then return a string.
Reply all
Reply to author
Forward
0 new messages