[jira] [Created] (MNG-7144) Spurious "Read timed out" errors during Maven dependency resolution

["Dominik Kaspar (Jira)" <ji...@apache.org>]

Dominik Kaspar created MNG-7144:
-----------------------------------

             Summary: Spurious "Read timed out" errors during Maven dependency resolution
                 Key: MNG-7144
                 URL: https://issues.apache.org/jira/browse/MNG-7144
             Project: Maven
          Issue Type: Bug
    Affects Versions: 3.6.0
            Reporter: Dominik Kaspar


Hello

We have hundreds of Maven builds each day, accessing a Nexus instance for dependency resolution. While this works very well, we rarely observe spurious download failures like this:
{noformat}
03:23:41 Downloading: https://company/nexus/content/groups/nexus-repo/org/elasticsearch/elasticsearch/7.8.0/elasticsearch-7.8.0.jar

[... some more successful dependency downloads and 30 minutes idle waiting ...]

03:53:42 [ERROR] Failed to execute goal on project SomeProj: Could not resolve dependencies for project
somegroup:someproj:jar:1.0.0-SNAPSHOT: Could not transfer artifact org.elasticsearch:elasticsearch:jar:7.8.0
from/to nexus-repo (https://company/nexus/content/groups/nexus-repo): GET request of:
org/elasticsearch/elasticsearch/7.8.0/elasticsearch-7.8.0.jar from nexus-repo failed: Read timed out -> [Help 1]
{noformat}
Observations:
 * The "missing" artifact always exists in the Nexus repository, sometimes it has been there for years
 * The read timeout always lasts exactly 30 minutes, which corresponds to the default '{{maven.wagon.rto}}'
 * Entries in the Nexus request.log suggest a successful transfer of the file, with a 200 status code and a typically short elapsedTime (the 30156 ms in this example were rather slow):
{noformat}
10.0.29.133 - jenkins [20/Feb/2021:03:24:11 +0100] "GET /nexus/content/groups/nexus-repo/org/elasticsearch/elasticsearch/7.8.0/elasticsearch-7.8.0.jar
HTTP/1.1" 200 - 2916352 30156 "Apache-Maven/3.6.0 (Java 11.0.9.1; Linux 3.10.0-1160.11.1.el7.x86_64)"
{noformat}

 * While in the above example, Maven 3.6.0 was used, we've observed the same problem with other versions, including 3.0.4 and 3.5.4

While network problems cannot be ruled out, it seems odd that this happens so rarely. What other causes could there be for such read timeouts? I'm out of ideas of how to tackle this issue...

Regards,
 Dominik



--
This message was sent by Atlassian Jira
(v8.3.4#803005)