Skip to main content

Linkiir Runtime Crashed

Symptoms​

  • /api/health reports Runtime as failed or degraded.
  • Workflows stop consuming messages.
  • Grid remains available but Runtime PID is absent.
  • The service supervisor repeatedly restarts Runtime.

Immediate actions​

  1. Record the first failure time.
  2. Confirm queue messages remain retained.
  3. Capture Runtime logs and operating-system events.
  4. Check memory, disk, file descriptors, and process limits.
  5. Identify the last workflow/node and message correlation ID before the crash.
  6. Disable a reproducibly crashing workflow only when necessary to stabilize the platform.
  7. Restart Runtime through Grid or the platform service, not by launching the binary manually.

Platform commands​

Windows:

Get-Service LinkiirGrid
Restart-Service LinkiirGrid
Get-ChildItem "C:\ProgramData\Linkiir\logs\runtime"

Linux:

systemctl status linkiir-grid
journalctl -u linkiir-grid --since "30 minutes ago"
sudo systemctl restart linkiir-grid

Docker:

./scripts/linkiirctl status
./scripts/linkiirctl logs linkiir-grid
./scripts/linkiirctl restart

Root-cause questions​

  • Did a specific script or native module trigger the crash?
  • Was the input unusually large or malformed?
  • Did Runtime exhaust memory or hit a timeout?
  • Did a dependency, schema, or shared library change?
  • Did the crash begin after an upgrade?

After recovery, verify consumer offsets, queue depth, message ordering, and downstream duplicates before resuming normal traffic.