App engine: I'm seeing 500 status and 204 error code

3
votes

I'm seeing a lot of requests with 500 response code. There are only one line on file log:

A problem was encountered with the process that handled this request, causing it to exit. This is likely to cause a new process to be used for the next request to your application. (Error code 204)

Also, I can see a weird appengine's auto log line. I'm using Python 2.7 with threading on and 5 idle instances:

190.210.97.5 - - [04/Sep/2015:10:00:21 -0700] "POST /api/v1/ecclaro/sondeos/event HTTP/1.1" 500 0 - "Java/1.6.0_24" "xxxxxxxxxxx.appspot.com" ms=4301 cpu_ms=0 pending_ms=942 exit_code=204 instance=00c61b117c96c7c9c0bec141a97c4d4f8ebb5189 app_engine_release=1.9.26

This happens more often when the traffic on my application is increasing. And often happen with some resources. I don't know to do. Because I can't see low level error logs. My assumption is it's a runtime failure of python but I don't know.

Anyone got any ideas?

Regards

1
google-app-enginegoogle-cloud-platform
Check and see if you instances are being restarted. You may have a memory problem. Normally you would see a soft memory error, but if you hit a big memory wall it could happen straight away. Also check if all occurrances are for the same transaction type. - Tim Hoffman
Thanks @TimHoffman. The answer is yes. I usually see these kinds of error. (soft memory errors). But I didn't notice that was the same problem. I've increased instance's tear from F1 to F2. Also I'll take a look my source code to looking for memory leaks. I hope it will enough. Thanks! - Nico
@nico did you ever solve this? We are also having this error and it doesn't appear to be memory related. - HJED
@HJED Hi! yes I managed to do it. In my case there wasn't a simple bug. This kind of problems appeared only when the thread accessed to my Cloud SQL instance. I had to improve my sql queries and I had to minimize the number of access to it. I suspect that the problem was a conectivity problem between app engine and Cloud SQL . - Nico
Ah, ok. Thanks, that gives me a better idea where to look. - HJED

1 Answers

2
votes

This issue and discussion suggest it is most likely an issue of running out of RAM.