fix(CubeNet): Remove dead code

[FakeLearne]

Remove dead code in /CubeNet

[cubesandboxbot]

Minor, non-blocking: ErrProgNotExist is an exported error in a library package (// Package cubevs is a library to manage CubeVS.). It has no in-repo references, so removal is fine, but it's technically a public API surface change — any external consumer of this package referencing the symbol would break. Acceptable for a dead-code cleanup; just worth being aware of. The two unexported removals (reapDNSLearnedPoliciesForInnerMap, ingressSession) are unambiguously safe, and keeping the ingressSessionValue struct (still used by the size assert in cubevs.go:327 and dump.go:466) is correct.

[cubesandboxbot]

Review: fix(CubeNet): Remove dead code (#1583)

This review was AI-generated.

Verdict: Approve. The PR deletes three symbols that are genuinely unreferenced across the entire repository, and the removals leave no orphaned imports or types. Clean, safe dead-code removal.

Verification

I searched the full base tree for each removed symbol:

Removed symbol	References outside definition	Verdict
ErrProgNotExist (cubevs.go)	None (definition + comment only)	Dead ✓
reapDNSLearnedPoliciesForInnerMap (dns_reaper.go)	None	Dead ✓
ingressSession (reaper.go)	None (already carried //nolint:unused)	Dead ✓

Import safety — no import becomes unused after the deletions:

• dns_reaper.go: fmt, errors, and ebpf all still have live call sites (reapDNSLearnedPoliciesForIfindex, reapDNSQueryTrack, etc.).
• reaper.go: fmt, uint32ToIP, ntohs remain used by egressSession (reaper.go:204-219).
• cubevs.go: errors remains used by ErrTooLong.

Type retention is correct — the PR keeps ingressSessionValue (reaper.go:221), which is the right call: the struct is still consumed by the static size assert at cubevs.go:327 and by the dump path at dump.go:466. Only the formatting helper was dead.

Minor note (non-blocking)

ErrProgNotExist is an exported identifier in a library package (// Package cubevs is a library to manage CubeVS.). Within the repo it has zero references, so this is fine to remove, but it is technically a public API surface change — external consumers compiling against this package would break. Given it's unused in-tree and the commit is framed as dead-code cleanup, this is acceptable; just flagging in case any downstream vendoring depends on it.

Suggested follow-up (optional)

Nothing blocking. If you want to keep the "dead code" cleanup complete, consider a quick go vet / build check on the CubeNet module — though static analysis shows no issues.