-
-
Notifications
You must be signed in to change notification settings - Fork 1.7k
Open
Description
Observed behavior
Under high-throughput WorkQueue workloads with frequent message acknowledgements, the filestore enters a state where it continuously logs warnings:
Cache lookup detected no cache: no idx present
This causes:
- Massive warning spam in logs
- High CPU usage
- High disk usage
The issue only resolves after server restart. No memory pressure observed.
Expected behavior
The server should handle high-throughput WorkQueue workloads without entering a degraded state that requires restart to recover.
Server and client version
nats-server v2.12.2
Host environment
Linux, observed in Kubernetes pods
Steps to reproduce
- Create a stream with WorkQueue retention policy
- Produce messages at high throughput
- Consume and acknowledge messages at high rate
- After running for some time, observe the warning spam begins
Metadata
Metadata
Assignees
Labels
No labels