Sleep pattern returns server error (500)

[voss]

Hi baby buddies,

I'm running Baby Buddy on a Linode "Nanode" instance, and it's been working flawlessly since April. I've installed it following the instructions in the install guide from github, for a manual deployment.

The Nanode is running Debian 10 (Buster), and has, as mentioned worked flawlessly since April when I installed it.

I haven't touched the config of either Baby Buddy or Debian in that timeframe, but I'm now no longer able to load the sleep pattern. When loading example.com/children/child/reports/sleep/pattern I get a white page that says:

Server error (500)

I checked the nginx error_log in /var/log/nginx but that is empty (which I realise may be because I haven't configured nginx properly).

I took a look at /var/log/uwsgi/app/babybuddy.log for any app errors, and I see entries like this, when I get the above mentioned error page:

[pid: 11720|app: 0|req: 39459/65061] 94.131.131.16 () {50 vars in 1100 bytes} [Fri Nov 19 11:16:27 2021] GET /children/child/reports/sleep/pattern/ => generated 145 bytes in 258 msecs (HTTP/1.1 500) 7 headers in 223 bytes (1 switches on core 0)

Am I looking in the right place for any errors, or do Baby Buddy log errors somewhere else as well?

[cdubz]

The easiest way to get detailed error information is to enable debug mode and try to the load graph. You can do by setting a DEBUG environment variable to True (from an .env file if you're using one or just at CLI) or just setting the value to True in settings/base.py#L25. Restart the services after that change and then try to load the graph again -- you should get much more detailed error output.

What version of Baby Buddy are you running? There have been a few bug fixes related to graphs recently so you may also want to try updating to the latest version as well.

[voss]

Thank you for your very fast reply!

I'm running version: v1.5.1 (4df001c). Good shout on potentially updating to the latest version. Can I go straight to the latest, 1.9.1 release, or should I update in steps to some of the previous updates first? And also, on that, is there a guide on how not to nuke your install when updating manually? (:

I will change DEBUG to True in the base.py file, and see what error output I get. Will the error output be sent directly to the error page, or should I still look in the babybuddy.log file?

Thanks a lot!

[cdubz]

You should be able to up the latest version fine, just make sure to run migrations after updating. And definitely make a database backup beforehand just in case.

Debug output will go to browser.

[voss]

Thanks for your help @cdubz.

I tried git pull, and was told that settings/base.py and urls.py had local changes, and had to be stashed or committed. I stashed them, ran git pull, and then git stash pop to merge those locally changed files. I guess merging that 1.5.1 base.py with the one from 1.9.1 would screw things up. I don't remember changing anything in urls.py though. After running git stash pop I then tried to run migrate which gave me this error:

/var/www/babybuddy/public:> python manage.py migrate
Traceback (most recent call last):
  File "manage.py", line 4, in <module>
    from dotenv import load_dotenv, find_dotenv
ImportError: No module named dotenv

After restarting uwsgi and nginx, I was met with a white page saying "internal server error" when loading, so I quickly rolled back as I was under some pressure to make this excellent app work again (:

I'm sorry for being so helpless here, but do you know if there's something else I should have done before running python manage.py migrate, or would this come down to me stashing the old base.py file?

After rolling back to 1.5.1, I enabled the debugging mode, and when trying to view the sleep pattern chart, I now get a much more verbatim error message which has this at the core:

Exception Type: TypeError
Exception Value: object of type 'float' has no len()
Exception location: ./reports/graphs/sleep_pattern.py, line 100, in sleep_pattern

I will try doing another upgrade later in the week where I don't stash anything, that might fix the error message I got from running migrate. Upgrading to 1.9.1 definitely seems like an easier solution, if I can get the update to run properly.

[cdubz]

Ah you pretty far back so you'll need to re-run pipenv install --three as well to make sure you have all the latest python dependencies. Then running migrations should work (side note -- you should tun the migrations through pipenv so pipenv run python manage.py migrate to ensure you use the environment packages.

If you can share the diffs for the settings and urls files I can possible also help you get that a bit better sorted, if necessary.

[voss]

Perfect. I think I will just try the update, now that I understand the process a bit better.

I was thinking of making a copy of the /var/www/babybuddy folder, and then update the app for that folder, to see if I can do it without any errors. My question is, updating a copy of the app, in that folder, wouldn't affect the instance currently running, as I'm only updating the virtual environment for that copy. Is that correct?

Sorry if these are very stupid questions, I have very little experience with pip and virtual environments (:

[cdubz]

No worries. Python's package management system is kind of shitshow so there are no stupid questions (:

Copying the folder as a backup should work because it should have the .git and .venv (where virtual environment libraries live) folders. That said, I'd recommend just creating a backup of the instance that you can fully restore from if something goes wrong (assuming that's possible with Linode -- I have never used it). Also, in theory, if something did get hosed you should still be able to checkout v1.4.1 and re-run pipenv install --three to get the correct packages back in place.

Also you may need to want to export PIPENV_VENV_IN_PROJECT=1 before any pipenv install commands. I don't think its necessary after the first time but I'm not 100% sure on that.

So my recommendation here is:

1. Backup Linode instance or tar -cxf ~/babybuddy.tar.gz /var/www/babybuddy
2. cd /var/www/babybuddy
3. git add --all && git stash
4. git fetch && git checkout v1.9.1
5. git stash pop
6. export PIPENV_VENV_IN_PROJECT=1 (just in case)
7. pipenv install --three
8. pipenv python manage.py migrate
9. sudo service uwsgi restart

And if you run in to irreparable issues anywhere along the way -- just restore the Linode instance backup or unpack the directory contents.

[voss]

Thank you so much for your patience and help @cdubz, this worked a treat!

Updated to 1.9.1. and now have sleep pattern again, and all the other new shiny things (: