You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
This repository has been archived by the owner on Jan 4, 2020. It is now read-only.
Our CI can sometimes exhibit some flaky behavior in that it'll fail for no good reason, and a retrigger solves the problem. Let's investigate what's going wrong (maybe it's a caching thing, or some state that isn't properly cleaned up, or some race condition?) and fix it!
Interestingly, the flakiness is different on BuddyBuild vs Travis.
Travis
Here's what a failed build log usually looks like on Travis:
Test Suite 'All tests' started at 08:31:26.018
Test Suite 'debug.xctest' started at 08:31:26.042
Test Suite 'MarathonTests' started at 08:31:26.042
Test Case 'MarathonTests.testInvalidCommandThrows' started at 08:31:26.042
Test Case 'MarathonTests.testInvalidCommandThrows' passed (0.002 seconds)
Test Case 'MarathonTests.testAddingAndRemovingRemotePackage' started at 08:31:26.044
Test Case 'MarathonTests.testAddingAndRemovingRemotePackage' passed (1.959 seconds)
Test Case 'MarathonTests.testAddingAndRemovingLocalPackage' started at 08:31:28.004
Test Case 'MarathonTests.testAddingAndRemovingLocalPackage' passed (0.729 seconds)
Test Case 'MarathonTests.testRemovingAllPackages' started at 08:31:28.733
Test Case 'MarathonTests.testRemovingAllPackages' passed (9.061 seconds)
Test Case 'MarathonTests.testAddingLocalPackage' started at 08:31:37.794
Test Case 'MarathonTests.testAddingLocalPackage' passed (0.54 seconds)
Test Case 'MarathonTests.testAddingLocalPackageWithDependency' started at 08:31:38.334
Test Case 'MarathonTests.testAddingLocalPackageWithDependency' passed (1.286 seconds)
Test Case 'MarathonTests.testAddingLocalPackageWithUnsortedVersionsContainingLetters' started at 08:31:39.620
Test Case 'MarathonTests.testAddingLocalPackageWithUnsortedVersionsContainingLetters' passed (0.914 seconds)
Test Case 'MarathonTests.testAddingAlreadyAddedPackageThrows' started at 08:31:40.534
Test Case 'MarathonTests.testAddingAlreadyAddedPackageThrows' passed (5.021 seconds)
Test Case 'MarathonTests.testTreatingNestedDependenciesAsAdded' started at 08:31:45.555
Test Case 'MarathonTests.testTreatingNestedDependenciesAsAdded' passed (2.605 seconds)
Test Case 'MarathonTests.testRunningScriptWithoutPathThrows' started at 08:31:48.160
Test Case 'MarathonTests.testRunningScriptWithoutPathThrows' passed (0.004 seconds)
Test Case 'MarathonTests.testRunningScript' started at 08:31:48.164
Test Case 'MarathonTests.testRunningScript' passed (3.542 seconds)
Test Case 'MarathonTests.testRunningScriptWithNewDependency' started at 08:31:51.705
travis_time:end:0d599c36:start=1496651472166114307,finish=1496651515992986208,duration=43826871901
The command "swift test" exited with 1.�
Done. Your build exited with 1.
As you can see above, it just exits with code 1, no additional debug information seems to be found 😢 It also seems like it stops on different test cases every time this happens, so don't think it's a problem with a single test case. Perhaps a timeout issue?
BuddyBuild
On BuddyBuild things are a lot more consistent, and should be easier to fix. It always fails on this test case:
Most of the time when I was getting strange errors in different CIs it was regarding timeouts. We're using free plans, so the bandwidth etc. is relatively slow.
Marathon test suite is heavy, so I wouldn't be surprised if it will turn out to be timeout errors.
Regarding buddybuild error. Maybe extracting try? folder.subfolder(named: "Packages").file(named: "Files").read() to a let can will help?
Feels like Swift 3.2 introduced new problem. Take a look at this integration .
I've faced this issue locally as well when I was running tests. It's stuck on testAddingAndRemovingRemotePackage or testAddingLocalPackage.
Our CI can sometimes exhibit some flaky behavior in that it'll fail for no good reason, and a retrigger solves the problem. Let's investigate what's going wrong (maybe it's a caching thing, or some state that isn't properly cleaned up, or some race condition?) and fix it!
Interestingly, the flakiness is different on BuddyBuild vs Travis.
Travis
Here's what a failed build log usually looks like on Travis:
As you can see above, it just exits with
code 1
, no additional debug information seems to be found 😢 It also seems like it stops on different test cases every time this happens, so don't think it's a problem with a single test case. Perhaps a timeout issue?BuddyBuild
On BuddyBuild things are a lot more consistent, and should be easier to fix. It always fails on this test case:
Rebuilding without cache seems to fix the problem. Perhaps some cleanup not being made?
The text was updated successfully, but these errors were encountered: