-
-
Notifications
You must be signed in to change notification settings - Fork 1.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Incoming messages to the host without connections #5539
Comments
Can you provide us detailed instructions on how we could reproduce that would be helpful. |
Hi, We tried to reproduce the same behavior on our dev environment , alas we are actually unable to reproduce it. When it occurred, we had a microservice sending messages to a subscriber, this subscriber is part of a queue group. If we stop the subscriber, we observe the behavior we can see in the graph sent by Jing, the number of messages is about x10. When we subscribe back the number of message come back to a normal number. We test it several time with exactly the same behavior, alas as i said, since we move to new nodes we are not able to reproduce this behavior. |
@derekcollison I can't reproduce it right now, but we have a node with this issue and if there is some possibility to debug it there - can do it. |
So you have a node that is showing this behavior that has no client connections and no jetstream assets on that node, correct? |
Right now it has clients but also for sure has additional unroutable traffic (if we fix one issue we will have this node without clients). We do not use JS so yes this node has no jetstream. |
Possible to see if issue presents with latest pre-release candidate for v2.10.17? RC6? https://github.com/nats-io/nats-server/releases/tag/v2.10.17-RC.6 |
I am trying to reproduce this situation but without luck, so we have it only on production. |
Is production showing the issue now? |
Yes, it is production, and the issue is present. |
Can we schedule a call to take a look? Shoot me an email - derek@synadia.com |
Observed behavior
We have a NATS cluster with 3 nodes in different AZ and connecting clients also to the node in the same AZ.
We have 1 node without connections, and once we got the situation when this node has no connections but a lot of incoming "orphan" messages from the other nodes:
We tried to change "pool_size = -1" - cluster updated the configuration but it did not help to down-crease incoming traffic.
Routes before the pool_size parameter change
Routes after the change
Only node restart did the trick. Situation after restart (0 connections, ~0 incoming bytes, ~0 outgoing bytes):
Expected behavior
Remove the subscription route when there are no consumers to read messages.
Server and client version
Server version: 2.10.14
Client: -
Host environment
GKE host:
OS: Container-Optimized OS from Google
OS version: 109
Architecture: x86-64
CR: containerd
Steps to reproduce
In our case a few slow consumers had appeared in a cluster and incoming messages ran to the free node.
In logs we could see only messages like:
Logs
The text was updated successfully, but these errors were encountered: