search

okarras / EmpiRE-Analysis

A Sustainable Literature Review for Analyzing the State and Evolution of Empirical Research in Requirements Engineering using KG-EmpiRE.
MIT License
6 stars 2 forks source link

Binder launch issue #1

Closed arnim closed 9 months ago

arnim commented 9 months ago

The binder launch link in the readme currently does not result in a successful build and start.

Waiting for build to start...
Picked Git content provider.
Cloning into '/tmp/repo2dockerk3u67oki'...
HEAD is now at a8d199b Update for re24
Building conda environment for python=3.11
Using PythonBuildPack builder

CC @rgaiacs if this is not something systematic let's track that here.

rgaiacs commented 9 months ago

The Kubernetes building pod has not extra information:

{"message": "Picked Git content provider.\n"}
{"message": "Cloning into '/tmp/repo2dockerde0s0lo8'...\n", "phase": "fetching"}
{"message": "Updating files:  98% (83/84)\r", "phase": "fetching"}
{"message": "Updating files: 100% (84/84)\r", "phase": "fetching"}
{"message": "Updating files: 100% (84/84), done.\n", "phase": "fetching"}
{"message": "HEAD is now at a8d199b Update for re24\n", "phase": "fetching"}
{"message": "Building conda environment for python=3.11\n"}
{"message": "Using PythonBuildPack builder\n", "phase": "building"}
rgaiacs commented 9 months ago

Dear @okarras,

do you know a good of this repository that you are aware that was working with Binder?

rgaiacs commented 9 months ago

Locally, repo2docker EmpiRE-Analysis works

Picked Local content provider.
Using local repo EmpiRE-Analysis/.
Building conda environment for python=3.11
Using PythonBuildPack builder
Step 1/50 : FROM docker.io/library/buildpack-deps:bionic
...

@yuvipanda @minrk any idea why PythonBuildPack builder might be failing silent on GESIS cluster?

arnim commented 9 months ago

Seems not to be specific to the GESIS cluster https://binder.curvenote.dev/v2/gh/okarras/EmpiRE-Analysis/HEAD?labpath=%2Fempire-analysis.ipynb

arnim commented 9 months ago

.gitattributes seems to be the culprit. Once .gitattributes is removed the build starts just fine.

https://mybinder.org/v2/gh/arnim/EmpiRE-Analysis/arnim-patch-2

@okarras is .gitattributes needed? Can you modify it so that the build passes?

okarras commented 9 months ago

Dear @okarras,

do you know a good of this repository that you are aware that was working with Binder?

I checked the commit history: https://github.com/okarras/EmpiRE-Analysis/commits/master?before=2eb7478dba09a041e59e97abb5f8b00feaa7fd1d+35

The version before Starting preparation for RE24 Artifact Track worked on binder. After the 16th of January, I was mainly changing the readme file and other documentation.

okarras commented 9 months ago

Seems not to be specific to the GESIS cluster https://binder.curvenote.dev/v2/gh/okarras/EmpiRE-Analysis/HEAD?labpath=%2Fempire-analysis.ipynb

Yes it is also not working on the mybinder.org page with the same "error".

okarras commented 9 months ago

.gitattributes seems to be the culprit. Once .gitattributes is removed the build starts just fine.

https://mybinder.org/v2/gh/arnim/EmpiRE-Analysis/arnim-patch-2

@okarras is .gitattributes needed? Can you modify it so that the build passes?

I had to add the gitattributes due to the use of GIT LSF as the folder supplemantary materials contains an RDF dump of the ORKG which is to large for GItHub itself.

I could remove it, but it does not explain why it work 4 weeks ago with the .gitattributes.

i do not really now much about the .gitattributes file, one of our developers help me with it.

rgaiacs commented 9 months ago

git log -- .gitattributes returns

commit 704b1a3e68b7ff61627f1a36d3d9d704052fdc33
Author: Oliver Karras <oliver.karras@tib.eu>
Date:   Mon Apr 24 13:56:33 2023 +0200

    added rdf dump

and git log 704b1a3e68b7ff61627f1a36d3d9d704052fdc33 -n 2 returns

commit 704b1a3e68b7ff61627f1a36d3d9d704052fdc33
Author: Oliver Karras <oliver.karras@tib.eu>
Date:   Mon Apr 24 13:56:33 2023 +0200

    added rdf dump

commit e2178b0669e0e52aa4eb06fe3d8c2eafdbd830df
Author: Oliver Karras <oliver.karras@tib.eu>
Date:   Mon Apr 24 11:07:08 2023 +0200

    added supplementary material

I tried both commits and both works:

Before .gitattributes was added: https://mybinder.org/v2/gh/okarras/EmpiRE-Analysis/e2178b0669e0e52aa4eb06fe3d8c2eafdbd830df (Work!)

After .gitattributes was added: https://mybinder.org/v2/gh/okarras/EmpiRE-Analysis/704b1a3e68b7ff61627f1a36d3d9d704052fdc33 (Also work!)

arnim commented 9 months ago

Hi @okarras,

Just tested if b8631c0 (the last commit before Starting preparation for RE24 Artifact Track) works as you said. It seems to have the same issue.

👉 https://mybinder.org/v2/gh/okarras/EmpiRE-Analysis/b8631c0

okarras commented 9 months ago

I checked the previous versions. Test replication on other system for this version exists a build after it. the error occurs.

rgaiacs commented 9 months ago

Dear @okarras,

I don't think that .gitattributes is the only culprit, see https://github.com/okarras/EmpiRE-Analysis/issues/1#issuecomment-1954065388, but your recent commit removing .gitattributes (https://github.com/okarras/EmpiRE-Analysis/commit/fc11a3e46ac882892f65db757d9b40eea1dfbe08) had a successful build.

If you want to figure out what is the culprit that is working with .gitattributes to make the build fail on MyBinder, you can use Git's bisect.

Setup

$ git bisect start 2eb7478dba09a041e59e97abb5f8b00feaa7fd1d 704b1a3e68b7ff61627f1a36d3d9d704052fdc33
Bisecting: 29 revisions left to test after this (roughly 5 steps)
[b9ef5e873ac69ea7f00b87bf3d46ba6b7cd74b47] Merge branch 'master' of https://github.com/okarras/EmpiRE-Analysis

2eb7478dba09a041e59e97abb5f8b00feaa7fd1d is the last commit when the issue was reported and 704b1a3e68b7ff61627f1a36d3d9d704052fdc33 is where .gitattributes was created.

Loop

  1. Run git rev-parse HEAD to obtain the Git commit
  2. Visit https://mybinder.org/
    1. fill " GitHub repository name or URL" with https://github.com/okarras/EmpiRE-Analysis/
    2. fill "Git ref (branch, tag, or commit)" with the output of git rev-parse HEAD
    3. click on "launch"
  3. Wait to see if MyBinder fails during "Using PythonBuildPack builder".
  4. Record the result using git bisect good or git bisect bad

After 5 iterations, you will get a message like

123ABC678DEF789 is the first bad commit
commit 123ABC678DEF789
Author:
Date:

    Commit message

Possible first bad commit

Based on https://github.com/okarras/EmpiRE-Analysis/issues/1#issuecomment-1954143742, the first bad commit is 6e411f4c6dc5fdd4386d42072f388e2ec1425cd6.

git log -3 6e411f4c6dc5fdd4386d42072f388e2ec1425cd6 returns

commit 6e411f4c6dc5fdd4386d42072f388e2ec1425cd6
Author: Oliver Karras <oliver.karras@tib.eu>
Date:   Tue Jan 16 15:18:26 2024 +0100

    replication test on second system

commit 75608edbdf2e8f5e7369e91fb346184ba369915d
Author: okarras <me@oliver-karras.de>
Date:   Tue Jan 16 10:46:50 2024 +0100

    Test replication on other system

commit fec29e1bdbffa1b009507e201d42d5ad477ca438
Author: okarras <me@oliver-karras.de>
Date:   Tue Jan 16 10:29:49 2024 +0100

    test on surface

https://mybinder.org/v2/gh/okarras/EmpiRE-Analysis/6e411f4c6dc5fdd4386d42072f388e2ec1425cd6 FAILS!

https://mybinder.org/v2/gh/okarras/EmpiRE-Analysis/75608edbdf2e8f5e7369e91fb346184ba369915d works.

Conclusion

Based on the information on Git commit 6e411f4c6dc5fdd4386d42072f388e2ec1425cd6 , I cannot tell why MyBinder is failing.

okarras commented 9 months ago

Dear @rgaiacs and @arnim

thank you so much for your help!

I tried a lot in the evening but I cannot really figure out what the problem is. I decided to remove the .gitattributes files and the RDF dump. I will provide it with another solution.

I would never have thought of it without your help. Thank you very much again.

Best Olli