Details
-
Improvement
-
Resolution: Fixed
-
Major
-
13.10
-
None
-
Unknown
-
N/A
-
N/A
-
Description
When Maven Central's Nexus instance is extremely slow, the build can fail with a timeout when performing staging repository operations:
[INFO] [INFO] * Upload of locally staged artifacts finished. [INFO] [INFO] * Closing staging repository with ID "orgxwiki-1547". [INFO] [INFO] Waiting for operation to complete... [INFO] ............................................................................................................................................................................................................................................. [INFO] [WARNING] TIMEOUT after 902.7 s [INFO] [INFO] [ERROR] Rule failure while trying to close staging repository with ID "orgxwiki-1547". [INFO] [ERROR] [INFO] [ERROR] Nexus Staging Rules Failure Report [INFO] [ERROR] ================================== [INFO] [ERROR] [INFO] [ERROR] [INFO] [ERROR] Cleaning up local stage directory after a Rule failure during close of staging repositories: [] [INFO] [ERROR] * Deleting context d7fd3687d7c21.properties [INFO] [ERROR] Cleaning up remote stage repositories after a Rule failure during close of staging repositories: [] [INFO] [ERROR] * Not dropping failed staging repository with ID "orgxwiki-1547" (Rule failure during close of staging repositories: []). [INFO] [ERROR] Remote staging finished with a failure: Staging rules failure!
The problem seems to be known (https://issues.sonatype.org/browse/OSSRH-40476) and the only thing we can do about it is to increase the nexus-maven-plugin's stagingProgressTimeoutMinutes parameter (default 5 minutes), in order to avoid failing the build since the operation will eventually be successful, just that it's very slow due to the high load.
Our currently configured timeout is 15 minutes, which still seems to not be enough. On this particular instance, it took around 25 minutes for the staging repo to get to the closed state.
The suggestion coming from tmortagne is to go to 60 minutes, as it is still better than having the build fail.