chore(Jenkinsfile) introduce Maven client side caching

[dduportal]

This PR is a first "real life" test of the BOM builds using the S3 PVC-based Maven client side caching.

It uses the ~6Gb archive (which is partial cache created in #4667) as a first tentative(s?).

The goal is to evaluate it does not slow down the BOM as we want this cache at least as a protection layer to avoid breaking bom build when ACP start receiving HTTP/500 from artifactory.

Ref. jenkins-infra/helpdesk#4525

Testing done

Submitter checklist

• Make sure you are opening from a topic/feature/bugfix branch (right side) and not your main branch!
• Ensure that the pull request title represents the desired changelog entry
• Please describe what you did
• Link to relevant issues in GitHub or Jira
• Link to relevant pull requests, esp. upstream and downstream changes
• Ensure you have provided tests - that demonstrates feature works or fixes the issue

[dduportal]

First tentative: the archive restoration time takes between 5 to 13 min during the parallel stages (while it takes ~1min on the prep stage). It does not scale well in the current setup.

Trying a 2 step process instead of a direct tar read: 1. copy archive from S3 share to local empty dir 2. uncompress archive from local empty to local empty. The goal is to identify what is not scaling well: local node filesystem (so step 2. would be slow) or S3 accesses (step 1 is slow).

[dduportal]

Second retry: the copy from S3 to local system is unsustainable. Gotta try with the aws s3 command instead, to rule out (or in) the S3 mount driver. Copy from the mountpoint (S3 CSI driver) to local filesystem (emptydir on a local nvme) of all the parallel pct stages takes > 10 min. Canceled the build

[dduportal]

• A curl command from the S3 presigned URL takes around 3min, which means the 6 Gb is download at a rate of ~34 Mbits/s.
• Retrying with the archive of 1.2 Gb generated from wip: attempts at caching Maven local repo #4667 (comment) which is expected to take around 35-40s to retrieve, way more sustainable

[dduportal]

• A curl command from the S3 presigned URL takes around 3min, which means the 6 Gb is download at a rate of ~34 Mbits/s.

  • Retrying with the archive of 1.2 Gb generated from wip: attempts at caching Maven local repo #4667 (comment) which is expected to take around 35-40s to retrieve, way more sustainable

With 1.2 Gb archive: jenkins-infra/helpdesk#4525 (comment)

It's hard to conclude because

• The cache did not cover enough dependencies
• There are contention issues which are blocking all analysis (tar extractions on local FS goes seconds to multiple minutes!).