[TomPlanche]

Which issue does this PR address?

Closes #4638

Rationale

A client that loses the caller of its connect must still come back, like a client that is Disconnected. Without this fix, only an explicit connect() recovers it, and nothing in the SDK calls one.

What changed?

When a caller dropped a connect() while the reconnect sweep waited between two passes, the client stayed in Connecting with no slots. keepalive() restarts the links only for Disconnected or for a slot without a task, so the heartbeat never restarted them, and a request never dials.

Now keepalive() also restarts the links when the abandoned flag of the ConnectCoordinator is set, as @numinnex suggested in the review of #4569. The new connect() sees the flag, and its sweep disconnects the half-done state before it sweeps again. The keepalive_restarts() guard does not change, so an explicit disconnect or a logout stays final. The test from the issue reproduces the bug and now passes.

The cancel inside resume_or_sign_in, after install_bootstrap, is a different path and is not part of this PR.

Local Execution

• Passed: cargo fmt --all and cargo clippy -p iggy --all-features --all-targets -- -D warnings.

• Passed: the test from the issue, which failed on master with left: Connecting:

  $ cargo test -p iggy --lib -- given_a_sweep_owner_cancelled
  test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 760 filtered out
  

• Passed: cargo test -p iggy --lib, with 761 tests.

• Passed: pre-commit hooks with prek run.

AI Usage

1. Claude Code (Claude Opus 5.5) checked that no PRs opened/drafted was fixing this.

[github-actions]

Thanks for the PR. It is labeled S-waiting-on-review and queued for review.

If AI assistance was used to create this PR, you are required to run the team-review skill before opening it.

Slash commands (own line, regular comment) move it around the queue:

• /ready - back to S-waiting-on-review after addressing feedback
• /author - flip to S-waiting-on-author while you finish changes
• /request-review @user-or-team - request a reviewer
• /pin - exempt the PR from the stale bot, /unpin to undo

See CONTRIBUTING.md for details.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 84.81%. Comparing base (689436f) to head (f0cdbdb).

❌ Your patch check has failed because the patch coverage (0.00%) is below the target coverage (50.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@              Coverage Diff              @@
##             master    #4782       +/-   ##
=============================================
- Coverage     90.89%   84.81%    -6.09%     
+ Complexity     3878     3876        -2     
=============================================
  Files          1564     1504       -60     
  Lines        407450   297379   -110071     
  Branches     341308   231237   -110071     
=============================================
- Hits         370368   252229   -118139     
- Misses        29706    37402     +7696     
- Partials       7376     7748      +372     

Components	Coverage Δ
Rust Core	84.63% <100.00%> (-7.73%)	⬇️
Java SDK	79.38% <ø> (ø)
C# SDK	84.37% <ø> (+5.98%)	⬆️
Python SDK	86.52% <ø> (ø)
PHP SDK	92.23% <ø> (ø)
Node SDK	96.75% <ø> (ø)
Go SDK	71.93% <ø> (+0.12%)	⬆️
C++ SDK	86.84% <ø> (ø)

Files with missing lines	Coverage Δ
core/sdk/src/links/connect.rs	94.16% <100.00%> (+0.06%)	⬆️
core/sdk/src/links/replicas.rs	96.73% <100.00%> (+0.15%)	⬆️

... and 467 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.