Issues with Gatsby deployment after dependency upgrades via PR for jenkins-infra/stories

[krisstern]

Service(s)

infra.ci.jenkins.io, Other

Summary

For the PR jenkins-infra/stories#79, we are experiencing some Gatsby deployment issues as shown in https://ci.jenkins.io/job/Websites/job/stories/job/PR-79/15/console where we see the following in the stack trace:

00:36:44    NestedError: Cannot copy from `.cache/.eslintrc.json` to `/opt/build/cache/cwd  /.cache/.eslintrc.json`: Cannot create directory `/opt/build/cache/cwd/.cache`  : EACCES: permission denied, mkdir '/opt/build'
00:36:44    
00:36:44    - index.js:274 pMap.concurrency.concurrency
00:36:44      [adapters]/[cpy]/index.js:274:12
00:36:44    
00:36:44    - index.js:141 
00:36:44      [adapters]/[p-map]/index.js:141:20
00:36:44    
00:36:44    - Caused By: NestedError: Cannot create directory `/opt/build/cache/cwd/.cache    `: EACCES: permission denied, mkdir '/opt/build'
00:36:44    
00:36:44    - fs.js:61 
00:36:44      [adapters]/[cp-file]/fs.js:61:8
00:36:44    
00:36:44    - index.js:15 async cpFileAsync
00:36:44      [adapters]/[cp-file]/index.js:15:2
00:36:44    
00:36:44    - index.js:269 async pMap.concurrency.concurrency
00:36:44      [adapters]/[cpy]/index.js:269:6
00:36:44    
00:36:44    - index.js:141 
00:36:44      [adapters]/[p-map]/index.js:141:20
00:36:44    
00:36:44    - Caused By: Error: EACCES: permission denied, mkdir '/opt/build'
00:36:44    
00:36:44  
00:36:44  [Pipeline] }
00:36:44  [Pipeline] // withEnv
00:36:44  [Pipeline] }
00:36:44  [Pipeline] // stage
00:36:44  [Pipeline] stage
00:36:44  [Pipeline] { (Deploy PR to preview site)
00:36:45  Stage "Deploy PR to preview site" skipped due to earlier failure(s)
00:36:45  [Pipeline] getContext
00:36:45  [Pipeline] }
00:36:45  [Pipeline] // stage
00:36:45  [Pipeline] stage
00:36:45  [Pipeline] { (Build Production)
00:36:46  Stage "Build Production" skipped due to earlier failure(s)
00:36:46  [Pipeline] getContext
00:36:46  [Pipeline] }
00:36:46  [Pipeline] // stage
00:36:46  [Pipeline] stage
00:36:46  [Pipeline] { (Deploy Production)
00:36:47  Stage "Deploy Production" skipped due to earlier failure(s)
00:36:47  [Pipeline] getContext
00:36:47  [Pipeline] }
00:36:47  [Pipeline] // stage
00:36:47  [Pipeline] }
00:36:47  
00:36:47  [Pipeline] // ansiColor
00:36:47  [Pipeline] }
00:36:47  [Pipeline] // timeout
00:36:47  [Pipeline] }
00:36:48  [Pipeline] // withEnv
00:36:48  [Pipeline] }
00:36:48  [Pipeline] // withEnv
00:36:48  [Pipeline] }
00:36:48  [Pipeline] // node
00:36:48  [Pipeline] End of Pipeline
00:36:48  ERROR: script returned exit code 1

We have checked there does not seem to be any problem with the build step, which seems to suggest the issue is purely Gatsby deployment related as discussed in https://community.cloudflare.com/t/gatsby-pages-deployment-error/558278/6.

We would appreciate if someone could take a look and let us know if anything could be done to resolve this.

Reproduction steps

Cannot reproduce the failure with the following steps carried out locally:

1. npm run build
2. npm run serve

[dduportal]

Quick note (while triaging this issue): the directory /opt/build does not seem a sane place for building temporarily a website.

I rings a bell but I can't recall the solution. I would check if there is an environment variable (maybe already set in the infra.ci pipeline) which allows setting an alternative directory

[krisstern]

Sure, that sounds like a good idea!