fix(helm): use cluster DNS for the node agent
Merged
RS
2d ago+42+42+01234567890123456789-0-0-0123456789
What type of PR is this?
- bug
What this PR does / why we need it:
Set dnsPolicy: ClusterFirstWithHostNet on the Snapshot agent DaemonSet.
The agent uses host networking. Without this policy, it falls back to the node’s DNS configuration, which may not resolve Kubernetes Service names. On a Liqo provider, this prevents the agent from reaching the consumer metadata API through curie-api.dynamo.svc.cluster.local, blocking checkpoint capture.
Which issue(s) this PR fixes:
Fixes #
How was this tested?
- helm lint charts/snapshot passed.
- helm unittest charts/snapshot/ passed: 94 tests.
- Both new DNS regression tests fail against the original DaemonSet template.
- git diff --cached --check passed.
Special notes for your reviewer:
Covers both ordinary and remote metadata configurations. Host networking remains enabled.
The corrected chart has not yet been deployed for live capture/restore verification.
Does this PR introduce an API change?
NONE
Additional documentation, e.g. enhancement proposals, usage docs:
docs/operations/security.md documents cluster DNS for the host-networked agent.
Checklist
- Commits are signed off (git commit -s), per CONTRIBUTING.md (../CONTRIBUTING.md)
- make check test passes locally
- Documentation is updated where behavior changed