There are two approaches for your issue.
First of all, you can try to use caches, so you don't need to download
dependencies over and over again. There are also settings which might help,
see the system properties in WAGON-545 [1]. E.g. retries and timeouts.
But.
To me it sounds from your description that your internal repo mirror needs
some HA / scaling set up.
Refer to the guides from nexus [2] and artifactory [3] for more
information. Have nodes near your locations and configure the LBs to prefer
near nodes.
[1] - https://issues.apache.org/jira/browse/WAGON-545
[2] -
https://help.sonatype.com/repomanager3/planning-your-implementation/resiliency-and-high-availability/high-availability-clustering-%28legacy%29/configuring-nodes
[3] - https://www.jfrog.com/confluence/display/JFROG/High+Availability
On Thu, 29 Sept 2022, 20:23 David Karr, wrote:
> I work in a very large enterprise that uses a centralized intranet maven
> repository to get artifacts from. It often has load issues that result in
> builds failing with "failed to respond". The team that maintains it is
> working towards eventual mitigations for that, but it will be quite a while
> before that actually happens.
>
> What knobs or dials can I get to to make the connection to the remote
> repository more resilient? The error "failed to respond" sounds like a
> connection timeout, not a read timeout, but I can't tell. What property
> values can I override that would help here?
>
