Travis fails with "Your build exited with 86."

Travis suddenly fails with the following:

Worker information
hostname: fc0d617b-ff8d-4b9b-a926-e9a048937681@1.worker-com-77564c74fb-j7trz.gce-production-2
version: 6.2.22 https://github.com/travis-ci/worker/tree/858cb91994a513269f2fe9782c15fc113e966231
instance: travis-job-fd9f8db0-9e66-470e-b01c-9bd1328266b1 travis-ci-sardonyx-xenial-1593004276-4d46c6b3 (via amqp)
startup: 6.423674619s
/home/travis/.travis/job_stages: line 13: unexpected EOF while looking for matching `)'
/home/travis/.travis/job_stages: line 1075: syntax error: unexpected end of file
travis_run_setup_filter: command not found
...
travis_run_finish: command not found
Done. Your build exited with 86.

I’m getting the same error with an almost empty .travis.yml:

image

3 Likes

This looks to be global and failing all builds.

I hit the same starting last night.
Travis status page doesn’t show anything.

Is this being looked upon?

Yeah our builds also started failing on Saturday. The earliest I see is a cron build on Saturday night Pacific Time but it could be that the builds have been broken earlier than that.

/home/travis/.travis/job_stages: line 531: syntax error near unexpected token `}'
/home/travis/.travis/job_stages: line 531: `}'
travis_run_setup_filter: command not found
travis_run_configure: command not found
travis_run_prepare: command not found
travis_run_disable_sudo: command not found
travis_run_checkout: command not found
travis_run_export: command not found
travis_run_setup: command not found
travis_run_setup_casher: command not found
travis_run_setup_cache: command not found
travis_run_use_workspaces: command not found
travis_run_announce: command not found
travis_run_debug: command not found
travis_run_before_install: command not found
travis_run_install: command not found
travis_run_before_script: command not found
travis_run_script: command not found
travis_run_before_cache: command not found
travis_run_create_workspaces: command not found
travis_run_cache: command not found
travis_run_after_success: command not found
travis_run_after_failure: command not found
travis_run_after_script: command not found
travis_run_finish: command not found
1 Like

We are facing the same issue.

Nothing in the status page, tho:

https://www.traviscistatus.com/

I’ve reported this to “Travis-CI Support” (support@travis-ci.com) but other than the automated email confirmation there’s been no response yet. Also no reaction to tweets to @traviscistatus and @travisci.

1 Like

We came across the same problem and found a workaround.

We have encrypted environment variables configured in .travis.yml and some of them contain punctuations like {}:"/ in their values. (I’m not sure which one is precisely problematic)

Removing the environment variables fixed the problem.

I’m not sure if it applies to all cases but I hope it’s useful.

This wouldn’t be a workaround.
As pointed out in the discussion, this is seen even for an almost empty .travis.yml with just linux distribution.

dist: trusty
os: linux
python: "3.6"

before_install:
  - echo $TRAVIS_PULL_REQUEST_BRANCH; echo ${TRAVIS_PULL_REQUEST_SHA}

Deleting the environment variable which contains '{"envs":[{"name":"ve... does avoid this error. A work-around of base64 encoding the environment variable and decoding it in the deploy script seems to work:

(part of) bash script to set env var:

  travisEnvValue=`echo -n "${sqt}{\"envs\":[{\"name\":\"${ve... | base64 -w 0`
  echo "travisEnvName: ${travisEnvName}"
  echo "travisEnvValue: ${travisEnvValue}"
  travis env set ${travisEnvName} "${travisEnvValue}"

(part of) bash deploy script:

travisSecret="${environment}_xyz"
echo "travisSecret: ${!travisSecret}"
echo "decoding travisSecret..."
xyz=`echo "${!travisSecret}" | base64 --decode`

Thank you! It needs to be changed in GUI settings as well as mine had it there.

Using the Travis CLI to set environment variables with the travis env set myVar myValue command is just an alternative to using the GUI to do the same.

Yes, mine had env vars both using travis CLI (.travis.yml) and some in GUI directly and hence didn’t catch with an empty file.

Having the same issues, builds failing since Saturday. Any long-term solutions?

I am no longer seeing the issue. I restarted a failed build and it’s running now. Not sure if they did something to fix it.

Hey all,

We have been working on this case and deployed a fix that should be resolved this issue. Restarting related build/job would be effective.

Thank you for your patience and understanding.

Thanks.

@mustafa was it an error or was it intended? Due the solution being related to a certain type of escaped keys