What whas wrong with these instances?

[Ivo Bellin Salarin]

Hi all,

I am stuck on a curious problem.

My application is getting a AttributeError: 'module' object has no attribute 'users_id_token'.
(The ‘module’ is endpoints. You certainly know that there are several versions of endpoints, certains including users_id_token, others not.For this reason, the problem loooks very similar to a sys.path issue.)

The full stack is pretty similar to:

AttributeError: 'module' object has no attribute 'users_id_token'
at <module> (/base/data/home/apps/s~project/frontprocess:3-24.399581252568876895/server/authentication/custom_bearer.py:173)
at <module> (/base/data/home/apps/s~project/frontprocess:3-24.399581252568876895/server/authentication/authentication_manager.py:2)
at <module> (/base/data/home/apps/s~project/frontprocess:3-24.399581252568876895/server/base_handler.py:53)
at <module> (/base/data/home/apps/s~project/frontprocess:3-24.399581252568876895/server/content/community/community_task.py:8)
at <module> (/base/data/home/apps/s~project/frontprocess:3-24.399581252568876895/server/main.py:12)

What’s strange, with such stack, is its statistical distribution. In the last 10 days, more than 3200 instances have successfully processed more than 6.5M requests having a similar stack. On this set, 9 instances have failed processing about 5k requests (the 80% of the requests they processed)

The instances failing those requests were:

00c61b117cff0b098bf102c264f02fed721dc3914cb9b6f989b0febfea8a8609b3ecb07922
00c61b117c9db2974268ab5c147236fbe41640a3c1cf526ce93aa8072eb683cbef9c38c4740236
00c61b117cd8f7454ef9756c275a9e7b522c7a35920a462ef6770a99a29439ab3dcd434462dd05
00c61b117cf67bff1251a6baaef4d398923c6aef0cf668a24edf04c3628a2e0c5e8f042e6c335636
00c61b117cb382e1d3e301c42754c742977859cd66b4c62cd8caf1072611a4eb3bc6a7fbca56b001
00c61b117c575699b48ea66920ac9b04c7b928fb6d4ecac51c30db4489bb90d456e7d90567c2c4
00c61b117c4277b3ad1c978bbee348f7b57507b33d3213265f4a46f4224a17dbeeb892b1f9c172
00c61b117cb4fef51ceef5852aa20183d1539961c35dcd79077c24c2c88f04abe493d171e8b139
00c61b117cdd6c72dccb980afbb00d58229128beaf2166bfc1b307517492b74daeef26089b99

Well, those instances do not exist anymore. The issue is not happening anymore since 3/16/17 @ 1h17.

But, what did have in common all the instances above? Why were they failing?

​

[Adam (Cloud Platform Support)]

If you can provide a method to reproduce this issue I'd advise posting to the issue tracker. There aren't any reported issues with sys.path or endpoints of a similar nature to what you're describing, and it's not likely much info can be gleaned from an instance ID for a deleted instance.