Code: Select all
The job exceeded the maximum log length, and has been terminated.
Code: Select all
The job exceeded the maximum log length, and has been terminated.
The builds have been failing for some time. There is a packaging problem. Travis updated the version of OSX they use which Peterl94 said is the problem. It looks like now there is a problem with the cache, Travis still seems to be building dependencies when it fails.
Any volunteers to test it? I think it could be handy for all our builds - no need to scroll through tons of logs after a failure. We don't have any sensitive data in the logs, nor we need to keep it, so solution No. 1 is my pick.1) You can pipe the verbose output of your command to a file and display the last 500-1000 lines on build failure, e.g.2) If you cannot see the error in the last 500-1000 lines of logs, you could upload your entire log file to sites like chunk.io or transfer.sh like this:Code: Select all
script: - command ... > cmd_output.txt after_failure: - tail -n 500 cmd_output.txt
The curl command would give you back a chunk.io URL (e.g.http://chunk.io/f/1edae590c82d46deb930b4237828d809) where you can access your whole log file. Ensure you have no sensitive data in your build logs if you use this method. This is probably the best method if you need to preserve your entire log, but if you only really need to see the errors, method 1 is better.Code: Select all
script: - command ... > cmd_output.txt after_failure: - curl -T cmd_output.txt chunk.io
3) Lastly, if you do have sensitive data in the logs, you can upload your command's log output file to S3 using your S3 credentials and our artifacts addon. You can see the documentation on how to do this here: http://docs.travis-ci.com/user/uploading-artifacts/.