You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
feat: Add --fail-fast option to stop on first failure (pdsh -k compatibility) #94
Add a new --fail-fast option to bssh that stops execution immediately when any node fails (connection error or non-zero exit code). This is a prerequisite for pdsh compatibility mode (meta-issue #91).
Parent Issue
Part of #91 (pdsh compatibility mode) - Phase 1: New Options for Feature Parity
Background
pdsh provides -k option that causes the tool to fail fast on any error:
Connection failure to any host
Non-zero return code from any command
This is useful for:
Critical operations where partial execution is unacceptable
Deployment scripts where all nodes must succeed
Validation checks across clusters
Proposed Implementation
CLI Interface
# Default behavior (continue on failure)
bssh -H "node1,node2,node3""risky-command"# Continues even if node1 fails, reports all results at end# Fail-fast mode
bssh -H "node1,node2,node3" --fail-fast "risky-command"# Stops immediately if any node fails, cancels pending operations
Option Definition
#[arg( short = 'k', long = "fail-fast", help = "Stop execution immediately on first failure\nCancels pending commands when any node fails or returns non-zero")]pub fail_fast: bool,
Implementation Tasks
Add --fail-fast / -k option to Cli struct in src/cli.rs
Implement failure detection in executor
Add cancellation mechanism for pending/running tasks
Implement proper cleanup of cancelled SSH sessions
Report which node caused the failure
Add unit and integration tests
Update help text and examples
Acceptance Criteria
Connection failure to any host triggers immediate stop
Non-zero exit code from any command triggers immediate stop
Pending (not yet started) commands are cancelled
Running commands are terminated gracefully
Error message clearly indicates which node failed and why
Exit code reflects the failure appropriately
Works with all execution modes (TUI, stream, normal)
Technical Considerations
Need to implement cancellation tokens or similar mechanism
Should integrate with tokio's task cancellation
Consider using tokio::select! for early termination
May need to track which tasks have started vs pending
Interaction with Other Options
--fail-fast + --require-all-success: Both require all nodes to succeed, but fail-fast stops early
--fail-fast + --parallel N: Should cancel pending tasks in the semaphore queue
Notes
Short option -k does not conflict with existing bssh options
This complements the existing --require-all-success and --check-all-nodes options
Activity
feat: Add --fail-fast option to stop on first failure (pdsh -k compat…
feat: Add --fail-fast option to stop on first failure (pdsh -k compat…
feat: Add --fail-fast option to stop on first failure (pdsh -k compat…