Doc build suddenly started to fail every time: core dumped

[diningphil]

Details

• Read the Docs project URL: https://readthedocs.org/projects/pydgn/
• Build URL (if applicable): https://readthedocs.org/projects/pydgn/builds/16667589/
• Read the Docs username (if applicable): https://readthedocs.org/profiles/diningphil/

For some reason unknown to me, the readthedocs builds started failing from middle/late March. I have even tried to compile previous versions of my repo, e.g. v1.0.0, which compiled successfully in the past. Now all I am getting is the core dumped error below. Have you changed some memory limit in the main servers that build the docs?

I would be happy to not install all the libraries in the docs/requirements.txt file, but that seems a prerequisite to correctly build the documentation in the readthedocs

Alternatively, is it possible to upload the compiled files instead of triggering the online build?

Thanks for your help

Expected Result

The doc should correctly build as it is happening in local and used to happen in the past. Now even prior versions that used to build successfully show the same problem.

Actual Result

[stsewd]

the readthedocs builds started failing from middle/late March.

Hi, the first build from that projects is from 2 days ago https://readthedocs.org/projects/pydgn/builds/?page=2, did you re-create that project?

Have you changed some memory limit in the main servers that build the docs?

No that I'm aware of

Alternatively, is it possible to upload the compiled files instead of triggering the online build?

That isn't possible at the moment, you may want to subscribe to #1083

Locally I got this error

terminate called after throwing an instance of 'std::bad_alloc'
  what():  std::bad_alloc
Aborted (core dumped)

I'll try increasing the memory limits of your project

[pydgn]

Thanks for the prompt response and for the help. Yes, I tried to re-create the project from scratch.

Just out of curiosity, how were you able to reproduce it locally? Could you please link the steps here?
I'm only able to build the doc locally from my conda/venv environment, so it does not throw any kind of error.

Thanks again! Hopefully the memory limit increase will sort the issue out =)

[stsewd]

@pydgn looks like your project is already building on a large builder (7 GB of ram).

how were you able to reproduce it locally?

I did that from my local instance of RTD, but I just tried on my local system (16 GB of ram) with the same result

$ cd docs
$ pip install -r requirements.txt
$ sphinx-build . _build/html
Running Sphinx v4.4.0
building [mo]: targets for 0 po files that are out of date
building [html]: targets for 14 source files that are out of date
updating environment: [new config] 14 added, 0 changed, 0 removed
terminate called after throwing an instance of 'std::bad_alloc'                                                         
  what():  std::bad_alloc
[1]    98043 IOT instruction (core dumped)  sphinx-build . _build/html

$ sphinx-build . _build/html
Running Sphinx v4.4.0
building [mo]: targets for 0 po files that are out of date
building [html]: targets for 14 source files that are out of date
updating environment: [new config] 14 added, 0 changed, 0 removed
[1]    98168 segmentation fault (core dumped)  sphinx-build . _build/html                                               

So, this doesn't look like a problem with RTD, maybe a transitive dependency was updated recently

[diningphil]

I was able to reproduce the segfault outside of the conda environment, thanks.

Any suggestion on how to discover the transitive dependency you are talking about? I'm really losing my mind here =)

Thanks in advance!

[stsewd]

Any suggestion on how to discover the transitive dependency you are talking about? I'm really losing my mind here =)

You could have compared the previous installed dependencies from previous builds, but since the project was re-created, those aren't available anymore. Maybe you could get those from someone who hasn't updated their dependencies and where the build passes. You could use pip freeze to get them.

I'm closing this issue since isn't a problem with RTD.