500 Service Error - when try to save a testrun

[jehoff]

WARNING: do not publically report security issues in the bug tracker!
Ping us via email to coordinate the fix and disclosure of the problem!

Description of problem

generated a testcase. Want to create a testrun. If i push "save" button to store my testrun "500 Service Error" pops up.

Component (web, API, etc)

Kiwi Docker Image @ Debian 8.0 System

Version or commit hash (if applicable)

3.49

How often reproducible

always

Steps to Reproduce

1. Generate Testcase
2. Confirm Testcase
   3.Write Testrun
3. Save Testrun

Actual results

Error pops ub

Expected results

no error

Additional info

i can create testcases, create users.... but i cant save test run

[xbln]

Did you change default system test_case_status or test_case_run_status by admin interface (because you wanted to have status entries in your language)?

[atodorov]

@jehoff please retry with latest master and also give me a traceback of the failure (enable DEBUG=True in settings). Also let us know if you have edited any of the default objects in the DB, especially changing the names in another language like @xbln says.

[jehoff]

first off all, sorry for my late replay. i wasn´t in office this morning.

@xbln : i didnt changed test_case_status nor test_case_run_status

@atodorov : what do you mean by "latest master"? where do i find the traceback/log.file? DEBUG is activited. I didnt change/modify anything with language.

i tried to update the container with:

docker pull kiwitcms/kiwi

but i got this error:

Error response from daemon: pull access denied for kiwitcms/kiwi, repository does not exist or may require 'docker login'

Should i try to reinstall kiwi docker-image?

thx for your fast support!

[atodorov]

@atodorov : what do you mean by "latest master"? where do i find the traceback/log.file? DEBUG is activited. I didnt change/modify anything with language.

With DEBUG=True traceback should be printed in the browser instead getting a 500 error page.

docker pull kiwitcms/kiwi

Try rebuilding the docker image from the latest checkout from git.

[jehoff]

@atodorov : There is no difference in browser window (debug on/off). only "shown 500 Error Service".
I will try to rebuild the latest git version

[xbln]

I would always suggest to use the latest https://github.com/MrSenko/kiwi-docker docker image. For that repository you need a login. You can ask Alex for the credencials.

To use the latest "official" image, preferably on a standard environment like the digitalocean docker droplet, will help to find out bugs by isolating them from errors made by rebuilding the image on a home grown environment.

For error tracking it was very useful for me adding a phpmyadmin container to the installation.

[atodorov]

@jehoff when you have DEBUG=True in your settings Django will not display a 500 page but a traceback and an autogenerated page in the browser. The fact that you are not seeing this means your settings have not been changed. Inside the running container Python traceback should be printed in /var/log/httpd/ssl_error_log.

You can try running Kiwi TCMS locally and see if you can reproduce:
http://kiwitcms.readthedocs.io/en/latest/set_dev_env.html

[okainov]

Inside the running container Python traceback should be printed in /var/log/httpd/ssl_error_log.

Is there more user-friendly way to check the log? I'm not an expert in web-docker stuff, but for me it looks like overhead to go into docker container for the logs, I would expect them to be in real filesystem\in some mounted docker volume, which is accessible from outside of the container
And could you also specify this somewhere in the documentation, like in FAQ: "where to find log files?" Thanks

[xbln]

It is not too hard to step "inside" the running container. Just type docker exec -it <container name> bash. To find out names of your containers, type docker container ls.

You can also copy content from inside a running container to your host filesystem by typing docker cp <container_name>:<path> <dest_path> or the other way round docker cp <source_path> <container_name>:<path>.

[atodorov]

Is there more user-friendly way to check the log?

We're working on making this easier to detect programatically and be reported to us automatically via Sentry. When done you can switch to your own Sentry credentials and debug yourself.

The easiest way is to set DEBUG=True and have the traceback in the browser. This is the standard Django practice.

I'm not an expert in web-docker stuff, but for me it looks like overhead to go into docker container for the logs, I would expect them to be in real filesystem\in some mounted docker volume, which is accessible from outside of the container

You are using a Docker container to run software so it is understandable that logs and other files will be inside the docker container. Exposing these directories via volumes and mounting them on the host will probably work (you have to update your docker-compose.yml to configure that) but be aware this can have all sorts of issues with file ownership and permissions between the host and the container not to mention SELinux. Also this kind of mapping isn't universally applicable to everyone because we have no idea in what environment do you run your docker compose or even if you use compose at all.

And could you also specify this somewhere in the documentation, like in FAQ: "where to find log files?"

Will do, this is definitely a good suggestion.

[LZ-001]

I also met this issue when saving a testplan, following is the error info got from ssl_error_log:
"RSA certificate configured for : does NOT include an ID which matches the server name"
what should I do?