Closed paulofrazao closed 2 years ago
The bug is caused by this change in the Flux CLI so the release.sh
script needs to be updated.
Issue still persists when building a new version. If you follow the steps below, error will occur when executing test.sh
git clone https://github.com/weaveworks/flux2-openshift.git cd flux2-openshift vim LATEST_VERSION - change version number npm install make release make generate
./test.sh
Taking a look.
@paulofrazao what is the error you're seeing?
When executing the test.sh script, we get the error below.
Attached is the error log. 746434899.zip
@paulofrazao I'm not sure if we're looking at the same code but the test.sh script doesn't use kind but ekz.
The reason the custom-8g62d pod is failing is this:
2022-10-11T17:07:46.734677245Z stderr F Error: open db-704484487: permission denied
2022-10-11T17:07:46.734850632Z stderr F Usage:
2022-10-11T17:07:46.734863867Z stderr F opm registry serve [flags]
2022-10-11T17:07:46.734866321Z stderr F
2022-10-11T17:07:46.734868205Z stderr F Flags:
2022-10-11T17:07:46.734869888Z stderr F -d, --database string relative path to sqlite db (default "bundles.db")
2022-10-11T17:07:46.734871631Z stderr F --debug enable debug logging
2022-10-11T17:07:46.734873344Z stderr F -h, --help help for serve
2022-10-11T17:07:46.734874937Z stderr F -p, --port string port number to serve on (default "50051")
2022-10-11T17:07:46.734876981Z stderr F --skip-migrate do not attempt to migrate to the latest db revision when starting
2022-10-11T17:07:46.734878424Z stderr F -t, --termination-log string path to a container termination log file (default "/dev/termination-log")
2022-10-11T17:07:46.734879816Z stderr F --timeout-seconds string Timeout in seconds. This flag will be removed later. (default "infinite")
2022-10-11T17:07:46.734881099Z stderr F
2022-10-11T17:07:46.734882532Z stderr F Global Flags:
2022-10-11T17:07:46.734884255Z stderr F --skip-tls skip TLS certificate verification for container image registries while pulling bundles or index
2022-10-11T17:07:46.734885517Z stderr F
The script is modified to remove ekz and and add kind.
The reason the custom-8g62d pod is failing is this:
2022-10-11T17:07:46.734677245Z stderr F Error: open db-704484487: permission denied 2022-10-11T17:07:46.734850632Z stderr F Usage: 2022-10-11T17:07:46.734863867Z stderr F opm registry serve [flags] 2022-10-11T17:07:46.734866321Z stderr F 2022-10-11T17:07:46.734868205Z stderr F Flags: 2022-10-11T17:07:46.734869888Z stderr F -d, --database string relative path to sqlite db (default "bundles.db") 2022-10-11T17:07:46.734871631Z stderr F --debug enable debug logging 2022-10-11T17:07:46.734873344Z stderr F -h, --help help for serve 2022-10-11T17:07:46.734874937Z stderr F -p, --port string port number to serve on (default "50051") 2022-10-11T17:07:46.734876981Z stderr F --skip-migrate do not attempt to migrate to the latest db revision when starting 2022-10-11T17:07:46.734878424Z stderr F -t, --termination-log string path to a container termination log file (default "/dev/termination-log") 2022-10-11T17:07:46.734879816Z stderr F --timeout-seconds string Timeout in seconds. This flag will be removed later. (default "infinite") 2022-10-11T17:07:46.734881099Z stderr F 2022-10-11T17:07:46.734882532Z stderr F Global Flags: 2022-10-11T17:07:46.734884255Z stderr F --skip-tls skip TLS certificate verification for container image registries while pulling bundles or index 2022-10-11T17:07:46.734885517Z stderr F
I spoke with Chanwit, Paulo Gomes and Stefan about this and no one can figure out why this is happening.
I have no clue what ekz is and I have no bandwidth to understand and maintain this distribution. If we can't find someone with OpenShift knowledge to maintain this repo, let's accept defeat and delete Flux from OpenShift. I see no value in having a Flux version on OperatorHub which is months behind upstream and is affected by major CVEs.
@paulofrazao have we taken this to any of our RH contacts and gotten their input?
@sjavage given that this repo uses custom tools like Javascript generators and EKZ which is a project owned by @chanwit, I don’t see how RH could help fix this.
EKZ is just a left over.
We have been using KIND in the test since March.
We have been using KIND in the test since March.
It's really hard to come up with fixes to issues that one can't reproduce by cloning the repo and following documentation. Please update the docs.
We have been using KIND in the test since March.
But this says otherwise: https://github.com/weaveworks/flux2-openshift/blob/main/test.sh#L4
Please update the docs.
and the scripts. 😄
Since taking over from Chanwit, I just modify the test.sh file before execution. I will permanently change this once this issue is resolved.
@paulofrazao have we taken this to any of our RH contacts and gotten their input?
@paulofrazao have we taken this to any of our RH contacts and gotten their input?
I haven't take this up with RH as they were assisting with Marketplace setup and not OpertorHub. The process is different, but we still need this addressed first as we want both the Marketplace and OperatorHub to have the same version.
Fixed by #16
Trying to build new Operator and with the latest version of 0.34.0, I get the following error:
make release ./release.sh Flux version: 0.34.0 Generating the manifests using the built CLI ... Exporting gotk-components.yaml ... make: *** [Makefile:2: release] Error 1
I also tried version 0.33.0 with no success. I went back to version 0.32 and this worked fine.
Looks like there were some changes after version 0.32.0, that is causing this issue.
https://user-images.githubusercontent.com/45883050/190167468-7b64071c-7d91-4064-bc48-50dce4a1e7b3.mp4