port opentelemetry operator #358
Conversation
Alarm 1DescriptionState transition like: # mutated 0
apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
name: test-cluster
spec:
config: "receivers:\n otlp:\n protocols:\n grpc:\n http:\nprocessors:\n\
\ memory_limiter:\n check_interval: 1s\n limit_percentage: 75\n spike_limit_percentage:\
\ 15\n batch:\n send_batch_size: 10000\n timeout: 10s\n\nexporters:\n \
\ debug:\n\nservice:\n pipelines:\n traces:\n receivers: [otlp]\n \
\ processors: []\n exporters: [debug]"
---
# mutated 1
apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
name: test-cluster
spec:
config: "receivers:\n otlp:\n protocols:\n grpc:\n http:\nprocessors:\n\
\ memory_limiter:\n check_interval: 1s\n limit_percentage: 75\n spike_limit_percentage:\
\ 15\n batch:\n send_batch_size: 10000\n timeout: 10s\n\nexporters:\n \
\ debug:\n\nservice:\n pipelines:\n traces:\n receivers: [otlp]\n \
\ processors: []\n exporters: [debug]"
podAnnotations:
actokey: actokey
---
# mutated 2
apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
name: test-cluster
spec:
config: "receivers:\n otlp:\n protocols:\n grpc:\n http:\nprocessors:\n\
\ memory_limiter:\n check_interval: 1s\n limit_percentage: 75\n spike_limit_percentage:\
\ 15\n batch:\n send_batch_size: 10000\n timeout: 10s\n\nexporters:\n \
\ debug:\n\nservice:\n pipelines:\n traces:\n receivers: [otlp]\n \
\ processors: []\n exporters: [debug]"
podAnnotations: {}
---
# mutated 3
apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
name: test-cluster
spec:
config: "receivers:\n otlp:\n protocols:\n grpc:\n http:\nprocessors:\n\
\ memory_limiter:\n check_interval: 1s\n limit_percentage: 75\n spike_limit_percentage:\
\ 15\n batch:\n send_batch_size: 10000\n timeout: 10s\n\nexporters:\n \
\ debug:\n\nservice:\n pipelines:\n traces:\n receivers: [otlp]\n \
\ processors: []\n exporters: [debug]"
podAnnotations:
actokey2: actokey2In the state ExplanationIt's a true alarm. When logging the variables related to the annotations in the operator (like FixCheck and fix the usage of k8s controller-runtime API to ensure the system's status is changed as the variables in the operator. Alarm 2DescriptionState transition like: # mutated 0
apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
name: test-cluster
spec:
config: "receivers:\n otlp:\n protocols:\n grpc:\n http:\nprocessors:\n\
\ memory_limiter:\n check_interval: 1s\n limit_percentage: 75\n spike_limit_percentage:\
\ 15\n batch:\n send_batch_size: 10000\n timeout: 10s\n\nexporters:\n \
\ debug:\n\nservice:\n pipelines:\n traces:\n receivers: [otlp]\n \
\ processors: []\n exporters: [debug]"
---
# mutated 1
apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
name: test-cluster
spec:
config: "receivers:\n otlp:\n protocols:\n grpc:\n http:\nprocessors:\n\
\ memory_limiter:\n check_interval: 1s\n limit_percentage: 75\n spike_limit_percentage:\
\ 15\n batch:\n send_batch_size: 10000\n timeout: 10s\n\nexporters:\n \
\ debug:\n\nservice:\n pipelines:\n traces:\n receivers: [otlp]\n \
\ processors: []\n exporters: [debug]"
image: actokey
ExplanationIt's a false alarm. The fake image can't be pulled thus the system will be in an error state. FixLet Acto generate data closer to real ones. It can improve the test coverage. Alarm 3DescriptionActo throws many alarms about the health status, but these alarms can not be steadily reproduced. ExplanationThese are false alarms. The pulled image is not included in preload images in FixTo fix this problem, let Acto can collect the correct initiate state. We can also let acto throw errors when finding that the tested cluster is pulling images, thus the users can manually fix the context. |
|
Hi @THUzxj , could you write a bug report for the first alarm in this bug template? https://github.com/open-telemetry/opentelemetry-operator/issues/new?assignees=&labels=bug%2Cneeds+triage&projects=&template=bug_report.yaml You can then file this bug to the official developer and get some credit:) |
|
The bug is cool. Would love to see the bug report. |
|
The bug report has been submitted as above. The new commit fixes the missed preload_image manually and marks some field as "required" manually for deeper testing. |
* port opentelemetry operator * Add the missed preload_images and required field
Porting opentelemetry operator to acto.