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
Describe the bug
My Kubernetes workers become unresponsive after a while, similar to this issue #2598. Unfortunately, I am unable to figure out why. While trying to do so I noticed Kubernetes workers have their memory accumulated over time, eventually triggering the OOM Killer.
Here are the logs from the Kubernetes worker before getting terminated. They are a pair of "connection closed" and "onnection successfully authorized" for one test session
Error dialing the worker: failed to WebSocket dial: failed to send handshake request: Get "http://k8s-worker-public-domain.com:9202/v1/proxy": EOF
Error dialing the worker: failed to WebSocket dial: failed to send handshake request: Get "http://k8s-worker-public-domain.com:9202/v1/proxy": EOF
Error dialing the worker: failed to WebSocket dial: failed to send handshake request: Get "http://k8s-worker-public-domain.com:9202/v1/proxy": read tcp 172.26.138.103:45962->1.234.567.89:9202: read: connection reset by peer
Error dialing the worker: failed to WebSocket dial: failed to send handshake request: Get "http://k8s-worker-public-domain.com:9202/v1/proxy": read tcp 172.26.138.103:45960->1.234.567.89:9202: read: connection reset by peer
Error dialing the worker: failed to WebSocket dial: failed to send handshake request: Get "http://k8s-worker-public-domain.com:9202/v1/proxy": EOF
error reading handshake result: failed to read protobuf message: failed to get reader: failed to read frame header: EOF
Session credentials were not accepted, or session is unauthorized
Session credentials were not accepted, or session is unauthorized
error reading handshake result: failed to read protobuf message: failed to get reader: failed to read frame header: read tcp 172.26.138.103:40810->1.234.567.89:9202: read: connection reset by peer
Error dialing the worker: failed to WebSocket dial: failed to send handshake request: Get "http://k8s-worker-public-domain.com:9202/v1/proxy": read tcp 172.26.138.103:40826->1.234.567.89:9202: read: connection reset by peer
To Reproduce
Steps to reproduce the behavior:
Connect to a Boundary target behind a Boundary Kubernetes worker
Notice the initial memory usage for the worker pod
Load-test the worker with a simple loop and keep it running.
After an hour or so, if the worker did not get OOMKilled, terminate the loop
Notice the pod's memory. You should find it relatively high than the initial memory, and it doesn't get lower with time.
In my case, the simple loop got the pod OOMKilled in 2 and half hours:
Describe the bug
My Kubernetes workers become unresponsive after a while, similar to this issue #2598. Unfortunately, I am unable to figure out why. While trying to do so I noticed Kubernetes workers have their memory accumulated over time, eventually triggering the OOM Killer.
Here are the logs from the Kubernetes worker before getting terminated. They are a pair of "connection closed" and "onnection successfully authorized" for one test session
The logs from Boundary CLI:
To Reproduce
Steps to reproduce the behavior:
In my case, the simple loop got the pod OOMKilled in 2 and half hours:
Expected behavior
Worker should free its unused memory after cancelling the session.
Additional context
Worker version: v0.15.4
Controller version: v0.14.3
CLI version: 0.14.2
The text was updated successfully, but these errors were encountered: