Add retries when downloading dependencies?

Question

Our CI runs failed because of a temporary Clojars blip:

Error building classpath. Could not transfer artifact
rewrite-clj:rewrite-clj:jar:1.1.45 from/to clojars
(https://repo.clojars.org/): Connect to repo.clojars.org:443
[repo.clojars.org/146.75.29.128] failed: Connect timed out 401 Error:
Process completed with exit code 1.

Does it seem like it could be good to add some retries to tools.deps when downloading dependencies to handle temporary failures?

Answer 1

401 means unauthorized. That doesn't seem like an error we would want to retry?

I'm not sure we can differentiate cases where retry would be worthwhile (and some of this is may be obscured through the Maven libs too).

Comments

The other aspect is that Maven may try multiple repos looking for the repo that holds the artifact. Adding retries would presumably also either take longer to make that search and/or add a lot of traffic to all repos.

---

We often get similar errors although not 401s in CI. Rerunning the build seems to always fix it.

Downloading: metosin/muuntaja/0.6.8/muuntaja-0.6.8.pom from clojars
...
Error building classpath. Failed to read artifact descriptor for metosin:muuntaja:jar:0.6.8
org.eclipse.aether.resolution.ArtifactDescriptorException: Failed to read artifact descriptor for metosin:muuntaja:jar:0.6.8
    at org.apache.maven.repository.internal.DefaultArtifactDescriptorReader.loadPom(DefaultArtifactDescriptorReader.java:259)
    at org.apache.maven.repository.internal.DefaultArtifactDescriptorReader.readArtifactDescriptor(DefaultArtifactDescriptorReader.java:175)
    at org.eclipse.aether.internal.impl.DefaultRepositorySystem.readArtifactDescriptor(DefaultRepositorySystem.java:255)
    at clojure.tools.deps.alpha.extensions.maven$fn__1061.invokeStatic(maven.clj:132)
    at clojure.tools.deps.alpha.extensions.maven$fn__1061.invoke(maven.clj:122)
    at clojure.lang.MultiFn.invoke(MultiFn.java:244)
    at clojure.tools.deps.alpha$expand_deps$children_task__754$fn__756$fn__757.invoke(alpha.clj:406)
    at clojure.tools.deps.alpha.util.concurrent$submit_task$task__479.invoke(concurrent.clj:35)
    at clojure.lang.AFn.call(AFn.java:18)
    at java.base/java.util.concurrent.FutureTask.run(FutureTask.java:264)
    at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1128)
    at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:628)
    at java.base/java.lang.Thread.run(Thread.java:829)
Caused by: org.eclipse.aether.resolution.ArtifactResolutionException: Could not transfer artifact metosin:muuntaja:pom:0.6.8 from/to clojars (https://repo.clojars.org/): /ci/.m2/repository/metosin/muuntaja/0.6.8/muuntaja-0.6.8.pom.part (No such file or directory)
    at org.eclipse.aether.internal.impl.DefaultArtifactResolver.resolve(DefaultArtifactResolver.java:425)
    at org.eclipse.aether.internal.impl.DefaultArtifactResolver.resolveArtifacts(DefaultArtifactResolver.java:229)
    at org.eclipse.aether.internal.impl.DefaultArtifactResolver.resolveArtifact(DefaultArtifactResolver.java:207)
    at org.apache.maven.repository.internal.DefaultArtifactDescriptorReader.loadPom(DefaultArtifactDescriptorReader.java:244)
    ... 12 more
Caused by: org.eclipse.aether.transfer.ArtifactTransferException: Could not transfer artifact metosin:muuntaja:pom:0.6.8 from/to clojars (https://repo.clojars.org/): /ci/.m2/repository/metosin/muuntaja/0.6.8/muuntaja-0.6.8.pom.part (No such file or directory)
    at org.eclipse.aether.connector.basic.ArtifactTransportListener.transferFailed(ArtifactTransportListener.java:52)
    at org.eclipse.aether.connector.basic.BasicRepositoryConnector$TaskRunner.run(BasicRepositoryConnector.java:369)
    at org.eclipse.aether.util.concurrency.RunnableErrorForwarder$1.run(RunnableErrorForwarder.java:75)
    at org.eclipse.aether.connector.basic.BasicRepositoryConnector$DirectExecutor.execute(BasicRepositoryConnector.java:628)
    at org.eclipse.aether.connector.basic.BasicRepositoryConnector.get(BasicRepositoryConnector.java:262)
    at org.eclipse.aether.internal.impl.DefaultArtifactResolver.performDownloads(DefaultArtifactResolver.java:514)
    at org.eclipse.aether.internal.impl.DefaultArtifactResolver.resolve(DefaultArtifactResolver.java:402)
    ... 15 more
Caused by: java.io.FileNotFoundException: /ci/.m2/repository/metosin/muuntaja/0.6.8/muuntaja-0.6.8.pom.part (No such file or directory)
    at java.base/java.io.FileInputStream.open0(Native Method)
    at java.base/java.io.FileInputStream.open(FileInputStream.java:219)
    at java.base/java.io.FileInputStream.<init>(FileInputStream.java:157)
    at org.eclipse.aether.internal.impl.DefaultFileProcessor.copy(DefaultFileProcessor.java:163)
    at org.eclipse.aether.internal.impl.DefaultFileProcessor.copy(DefaultFileProcessor.java:151)
    at org.eclipse.aether.internal.impl.DefaultFileProcessor.move(DefaultFileProcessor.java:252)
    at org.eclipse.aether.connector.basic.BasicRepositoryConnector$GetTaskRunner.runTask(BasicRepositoryConnector.java:482)
    at org.eclipse.aether.connector.basic.BasicRepositoryConnector$TaskRunner.run(BasicRepositoryConnector.java:364)
    ... 20 more

---

the big stacktrace above is due to a missing local file, not a problem finding a network resource:

   Caused by: java.io.FileNotFoundException: /ci/.m2/repository/metosin/muuntaja/0.6.8/muuntaja-0.6.8.pom.part (No such file or directory)

hard to say with knowing more about the ci setup, but maybe some kind of problem with a shared /ci filesystem

---

I just went back and looked at the logs and 401 was the line number of the log that I had included when copying.

Here's the logs with timestamps:

00:21:14 GMT Downloading: org/clojure/data.codec/0.1.0/data.codec-0.1.0.jar from central
00:21:14 GMT Downloading: com/amazonaws/jmespath-java/1.11.713/jmespath-java-1.11.713.jar from central
00:21:14 GMT Downloading: clj-commons/pomegranate/1.2.1/pomegranate-1.2.1.jar from clojars
00:21:24 GMT Error building classpath. Could not transfer artifact rewrite-clj:rewrite-clj:jar:1.1.45 from/to clojars (https://repo.clojars.org/): Connect to repo.clojars.org:443 [repo.clojars.org/146.75.29.128] failed: Connect timed out
00:21:24 GMT Error: Process completed with exit code 1.

I think you might want to catch a ConnectTimeoutException if you were going to retry:  https://www.javadoc.io/doc/org.apache.httpcomponents/httpclient/4.3.4/org/apache/http/conn/ConnectTimeoutException.html

---

lol on the 401.

I'm open to this idea if we can distinguish in the right place between cases where the artifact doesn't exist vs where the service is unavailable. It's not clear to me that we have that opportunity and can tell that, so some more investigation is needed.