fluent-operator
fluent-operator copied to clipboard
bug: kubesphere/fluent-bit:v1.9.9 fluent-bit-watcher not executable
Describe the issue
Running kubesphere/fluent-bit:v1.9.9 gives output:
exec /fluent-bit/bin/fluent-bit-watcher
To Reproduce
docker run --rm -it kubesphere/fluent-bit:v1.9.9
of cause same when deployed via fluent-operator master in Kubernetes (kind)
Expected behavior
it runs, showing proper output similar to v1.9.8 :smile:
Your Environment
- Fluent Operator version: current master 91e1e98e0c08dd298e9f2dcf081ccbd20e6c27ea, but irrelevant here
- Container Runtime: Docker version 20.10.18, build b40c2f6
- Operating system: Ubuntu 20 LTS
- Kernel version: 5.15.0-48-generic
How did you install fluent operator?
No response
Additional context
No response
@wenchajun would you help to check if any of the changes introduced in v1.5.1 is related to this? If the official fluentbit:v1.9.9 has the same problem or the fluentbit image we made has the problem?
I thought the /fluent-bit/bin/fluent-bit-watcher is the extra piece you are adding to the official fluentbit image, no?
I thought the
/fluent-bit/bin/fluent-bit-watcheris the extra piece you are adding to the official fluentbit image, no?
Yes, the fluentbit in our custom image is started by /fluent-bit/bin/fluent-bit-watcher
just wanted to report I also got this. I remember before I didnt get this, so does anyone know what version we might need to roll back to in order to get it working again?
edit: nvm you said "it runs, showing proper output similar to v1.9.8" so that answers that question, anyways for thoroughness heres my output:
kubectl get pods -n fluent
NAME READY STATUS RESTARTS AGE
fluent-bit-bkzbj 0/1 CrashLoopBackOff 4 (46s ago) 2m21s
fluent-bit-kf5q5 0/1 CrashLoopBackOff 4 (53s ago) 2m21s
fluent-operator-749bfb89bd-kcc6m 1/1 Running 0 2m32s
username@hostname:~$ kubectl logs fluent-bit-bkzbj -n fluent
exec /fluent-bit/bin/fluent-bit-watcher: no such file or directory
@
just wanted to report I also got this. I remember before I didnt get this, so does anyone know what version we might need to roll back to in order to get it working again?
edit: nvm you said "it runs, showing proper output similar to v1.9.8" so that answers that question, anyways for thoroughness heres my output:
kubectl get pods -n fluent NAME READY STATUS RESTARTS AGE fluent-bit-bkzbj 0/1 CrashLoopBackOff 4 (46s ago) 2m21s fluent-bit-kf5q5 0/1 CrashLoopBackOff 4 (53s ago) 2m21s fluent-operator-749bfb89bd-kcc6m 1/1 Running 0 2m32s username@hostname:~$ kubectl logs fluent-bit-bkzbj -n fluent exec /fluent-bit/bin/fluent-bit-watcher: no such file or directory
@wenchajun Would you take a look if you can reproduce this
After my testing, it seems that the underlying image is the cause here, and the exact cause is still being troubleshot. You can switch the image to kubesphere/fluent-bit:v1.9.4 for now.
I've revert #416 by https://github.com/fluent/fluent-operator/pull/439 which might cause this problem and @wenchajun will release v1.6.1 @ITler @perezjasonr