Launchable is joining CloudBees! We’re combining forces to usher in the next generation of DevOps. Learn more.

Troubleshooting

"launchable verify" failure

Firewalls and static IP addresses

If you receive an error like this one, then you'll need to configure your firewall to allow traffic to api-static.mercury.launchableinc.com:

$ launchable verify
unable to post to https://api.mercury.launchableinc.com/...

$ launchable record build
...
Exception in thread "main" java.net.UnknownHostException: api.mercury.launchableinc.com: No address associated with hostname

If you need to interact with the API via static IPs, first set the LAUNCHABLE_BASE_URL environment variable to https://api-static.mercury.launchableinc.com.

The IP for this hostname will be either 13.248.185.38 or 76.223.54.162 which you can add to your firewall settings.

Proxies

If your CI server sits behind a proxy, you can tell the CLI to use it by setting the HTTP_PROXY and/or HTTPS_PROXY environment variables. For example:

export HTTP_PROXY="http://10.10.1.10:3128"
export HTTPS_PROXY="http://10.10.1.10:1080"

SSL certificate verification error

If you get an error like this:

SSLError(SSLCertVerificationError(1, '[SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: unable to get local issuer certificate

It is an indication that your system is lacking the root CA certificates. See the documentation for Requests, which the CLI uses under the hood, for how it looks up certificates. Also check out Stack Overflow posts like this where people discusses various remedies.

If all else fails, use the --skip-cert-verification option of the Launchable CLI to altogether bypass the SSL certificate check process. This means you are susceptible to MITM attacks, so use it with caution.

Missing Git object during commit collection

During record build or record commit , you might encounter errors like this:

Warning: 5730667 is missing. Skipping diff calculation for f2244ec -> 20630e5.

This warning indicates that a git object (typically a "blob", which stores a file) is missing from the local clone, and that prevented us from calculating a diff between the two commits listed. This is typically a result of Git shallow clone or Git partial clone.

Launchable will continue to function without this information, hence this is a warning; Predictive test selection will run with limited set of information, which will hurt the performance by unknown amount. If you are only seeing this sporadically, we suggest you simply ignore this warning.

Getting rid of this warning involves tweaking the git-clone or git-fetch operations to ensure enough of the relevant Git objects are made available locally. For example,

  • CI systems often use --depth 1 to just fetch the very latest commit and that alone. Increase this number should help, say --depth 32

  • GitHub Action actions/checkout uses --filter=blob:none when you choose to do a sparse checkout. Use filter: none to override this.

Recording branch of build

During record build, Launchable tries to automatically determine the Git branch being built. However, this can fail, depending on how you check out the code. When this happens, our webapp will ask you to configure this explicitly.

If your build involves just one repository, you can do this by using the --branch NAME option:

launchable record build --name $BUILD_NAME --source . --branch $GIT_BRANCH

If you have multiple repositories, you have to use the --branch REPONAME=BRANCHNAME format:

launchable record build --name $BUILD_NAME \
  --source product=. --source test=../test --branch product=$GIT_BRANCH

You can specify branches to every repositories in this manner, but the first repository is the most important, as that is used as the branch of this build as a whole. We recommend choosing the primary repository as. the first repository for this reason.

Integration Issues

When potential integration issues are detected in your test session, our web app displays an Integration Issues warning. Please follow the steps below to address the warning:

First, check whether you are using Zero Input Subset.

  • If you are not using it, go to the next question.

  • If you are using it, confirm whether you have specified the –test-suite option.

    • If you have not specified it, you should specify the –test-suite option. (See here for details.)

    • If you have specified it, no further action is needed.

Secondly, check the result after executing launchable record tests command. Are the numbers of “Files found” and “Tests found” correct?

  • If the numbers are not correct, Here are some possible solutions:

    • Check the regular expression in the launchable record tests command. It might not specify the correct path to report tests.

    • Check your test report files. They might not contain all test results.

  • If the numbers are correct, go to the next question.

Finally, check whether there are any tests that passed when requesting a subset but are not included in the test report files.

  • If no tests are missing, please contact us.

  • If some tests are missing, check whether they can be included in the test report files.

    • If they cannot be included, please contact us.

    • If they can be included, please update the report files accordingly.

Last modified ago