OOMKilled on Teams does not attaching the correct log for the container instead of it the logs attached is for the first container
antikilahdjs opened this issue · 4 comments
Describe the bug
Hello, I have many containers inside of pod then when the OOM occurs and sent it to Teams, the log attached is not for the container with the OOMkill instead of it the logs is for the first container. As example, I have the pod with these names below:
1 - pod-test-logs-oomkill
- container-test-01
- containerd-massive-text-01
- containerd-massive-text-02
If the OOMkill occurs in container containerd-massive-text-02 the logs attached is for the container-test-01 because the container is the first in the list.
The image below will demonstrate it
The container with the oomkilled was recorder-1026 but the logs attached was for the container rtsp-proxy
To Reproduce
Steps to reproduce the behavior:
1 - Install using the official helm charts
2 - Create a pod with 2 or more containers
3 - I have a Prometheus configured with Alertmanager
4 - Configure the SINK to use Teams
5 - My trigger inside of the helm is:
- triggers:
- on_pod_oom_killed:
rate_limit: 3600
actions:
- pod_oom_killer_enricher: {}
- logs_enricher: {}
- pod_node_graph_enricher:
resource_type: Memory
display_limits: true
- oomkilled_container_graph_enricher:
resource_type: Memory
display_limits: true
stop: true
Expected behavior
The logs attached needs to be for the container with the OOMkilled and not for the top 1 container in the list
Screenshots
It was added above
Desktop (please complete the following information):
- OS: RedHat 8.5 and Ubunut 20.04LTS
- Browser: Chrome
- Version: 119
Additional context
Add any other context about the problem here.
@antikilahdjs thanks for reporting.
Some thoughts on a fix: We should probably fetch the logs inside pod_oom_killer_enricher
as there we have context on which container was OOMKilled. Architecturally, logs_enricher
just doesn't have that information.
@antikilahdjs would you be interested in contributing a fix for this? Also very curious where those screenshots are from. Looks like an interesting frontend for Kubernetes.
Hello @aantn, thanks for your comments and I really sorry but I am not a developer and I cannot help with the code base to help on it. There is another way to fix it or really need a re-factore code?
So if you need to know about my frontend you can reach out here https://github.com/kubesphere/kubesphere
The fix is included in 0.10.26 which was released today