Clicking on resources tab gives 502 error sporadically

[linzjax]

Steps to reproduce the error

Go to a project you don't have permission to view resources on (this is the one I was looking at)
Click between the resource tab and the overview tab until you get a 502 error.

Actual behavior

Sporadic 502 errors.

Expected behavior

I should see an empty table every time.

[ian-ross]

Nasty. I wonder what that could be. 502s normally come from nginx or UWSGI. Maybe one for @amplifi?

[linzjax]

error log print out:

Subject[Django] ERROR (EXTERNAL IP): Internal Server Error: /organizations/olivers-org/projects/olivers-project/resources/
DateToday at 9:33 AM
Internal Server Error: /organizations/olivers-org/projects/olivers-project/resources/

UnicodeEncodeError at /organizations/olivers-org/projects/olivers-project/resources/
'ascii' codec can't encode character '\xea' in position 1207: ordinal not in range(128)

...
Traceback: 

File "/opt/cadasta/env/lib/python3.5/site-packages/django/template/base.py" in _resolve_lookup
883. current = current[bit]

File "/opt/cadasta/env/lib/python3.5/site-packages/django/template/context.py" in __getitem__
77. raise KeyError(key)

During handling of the above exception ('unarchive_url'), another exception occurred:

File "/opt/cadasta/env/lib/python3.5/site-packages/django/template/base.py" in _resolve_lookup
889. if isinstance(current, BaseContext) and getattr(type(current), bit):

During handling of the above exception (type object 'RequestContext' has no attribute 'unarchive_url'), another exception occurred:

File "/opt/cadasta/env/lib/python3.5/site-packages/django/template/base.py" in _resolve_lookup
898. current = current[int(bit)]

During handling of the above exception (invalid literal for int() with base 10: 'unarchive_url'), another exception occurred:

File "/opt/cadasta/env/lib/python3.5/site-packages/django/template/base.py" in _resolve_lookup
905. (bit, current)) # missing attribute

[ian-ross]

Oh no. It's the Unicode logging error, back to haunt us like the ghost of Jacob Marley.

[amplifi]

Confirmed; this error is bubbling up from Django. Everything server-side is still set to UTF-8.

[linzjax]

It might also be this, but since I can't reproduce the error I'm not sure. It's missing the unarchive_url definition that happens in this template

[ian-ross]

Hmmm. Should the form_modal block in project_list.html really have that modal in it at all? @seav added that -- is it needed there, Eugene? The resource unarchiving modal doesn't seem to be referenced anywhere within that template.

(None of this explains why we're getting these sporadic logging errors, but it might explain the increase in their frequency again.)

[linzjax]

So, it needs to be there in order for this button to work.

[ian-ross]

Hmmm. Eugene wrote some pretty ugly JS to do this, which is why it's been really hard to track down what's going on. The unarchive URL is set up from a HTML data attribute in the JS at the bottom of the resources/table.html template. When the resources/project_list.html template is rendered, including the resources/modal_unarchive.html template, unarchive_url isn't assigned, so I would have thought that the templating engine would complain at that point because it can't find unarchive_url in the template context. That's what seems to be happening on the production sites sometimes, but not in the development VM, and not 100% of the time. Any ideas, @seav?

[ian-ross]

@seav When I said "pretty ugly JS", that's not a pejorative judgment against you, by the way. The JS more or less always ends up being ugly in these cases, because you have lots of little bits of it lying around in different places, and it's hard to track down what's going on.

[seav]

Hmmm... I've never encountered any errors in the dev VM because of the lack of the unarchive_url variable, but if it helps to eliminate a possible cause for the 502 error, we can add a dummy unarchive_url to the include call:

{% include 'resources/modal_unarchive.html' with unarchive_url="#" %}

[ian-ross]

Let's try that and see what happens. I'll make a tiny PR for it, and we can merge it and ask @amplifi to deploy to staging today or tomorrow so we can see if it makes a difference. Since we don't have any other leads for why this is happening, trying this is as good an idea as any, I think.

[seav]

So far I haven't encountered any 502 errors while browsing the staging site, which should already have PR #797. Can anybody else confirm?

[linzjax]

I can't make any pop up. Are org admins now allowed to see deleted resources?

If they are it might be worth adding the same Active/Archived filter to the resources table. Definitely not urgent though.

[dpalomino]

Hi,

I haven't seen any other 502 errors either. I think we could close this issue and if needed open a new one for the active/archive resources (low prio)...

This is what I see for deleted resources (as a org admin):