Open ChaiBapchya opened 5 years ago
Hey, this is the MXNet Label Bot. Thank you for submitting the issue! I will try and suggest some labels so that the appropriate MXNet community members can help resolve it. Here are my recommended labels: Scala, Test, CI, Build
@mxnet-label-bot add [Scala, Test, CI, Build]
@ChaiBapchya This issue is that the network connection failed while downloading a maven dependency. I don't know how we handle this besides rerunning the jenkins job
Haven't looked at the code. But isn't there any way to catch exception and troubleshoot, instead of asking contributors to retriever CI?
The problem doesn't lie on Maven so I don't think there is anything Maven can do to address it directly. Usually we would add some amount of retries for downloading. I just looked and there doesn't seem to be a way to make Maven retry downloading. I would be hesitant to retry the entire Maven test because this doesn't seem to be a frequent problem and it would immediately multiply the cost of the Scala tests.
We could try to do something to improve the error message. Currently, it is:
Failed to execute goal on project mxnet-examples: Could not resolve dependencies for project org.apache.mxnet:mxnet-examples:jar:INTERNAL: Could not transfer artifact nu.pattern:opencv:jar:2.4.9-7 from/to central (https://repo.maven.apache.org/maven2): GET request of: nu/pattern/opencv/2.4.9-7/opencv-2.4.9-7.jar from central failed: Connection reset -> [Help 1]
I feel like this error message is clear enough that it was some kind of networking problem. Are you thinking some kind of CI specific error messaging?
@zachgk I just feel it is unreasonable to expect contributors/MXNet users to retrigger the PRs because something wasn't downloaded correctly.
Now specific to this issue - Error says GET request failed. Would be great if it adds way to solve it (currently retrigger CI, hopefully in future it auto-corrects itself)
But going forward, we need to make CI robust enough for connection failures.
@perdasilva Any idea why the CI might be having problems connecting to maven?
Scala unix GPU build error in an unrelated PR #15541
Pipeline - http://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/mxnet-validation%2Funix-gpu/detail/PR-15541/5/pipeline/