Closed agardnerIT closed 1 year ago
Seeing lots of this during install. Maybe temporary, may be important but it is a killercoda problem not Keptn:
E0124 23:13:02.297892 25726 memcache.go:106] couldn't get resource list for metrics.k8s.io/v1beta1: the server is currently unable to handle the request
E0124 23:13:02.325939 25726 memcache.go:255] couldn't get resource list for metrics.k8s.io/v1beta1: the server is currently unable to handle the request
E0124 23:13:02.367077 25726 memcache.go:106] couldn't get resource list for metrics.k8s.io/v1beta1: the server is currently unable to handle the request
E0124 23:13:02.423857 25726 memcache.go:255] couldn't get resource list for metrics.k8s.io/v1beta1: the server is currently unable to handle the request
Everything (eventually) starts up correctly:
This works:
This works:
This works:
This works:
This works (this time and for me):
Q on slack: The first thing I encountered was that it hangs for a LONG time at this step: ubuntu $ kubectl -n keptn-jes wait --for condition=Available=True --timeout=${POD_WAIT_TIMEOUT_MINS}m deployment/job-executor-service. It finally gets to "Installation Complete" after 11 minutes.
The Job Executor Service (JES) needs to communicate with Keptn so until all the pods in the keptn
namespace are running, the JES pod won't start.
Unfortunately the startup time has as much to do with the underlying infrastructure (provided for free by Killercoda) so there isn't much we can do here.
So what's happening?
A few possibilities I see:
1) The user is attempting to run setup_keptn.sh
on a preexisting repo. DO NOT DO THIS. Let the demo create a new repo every time you re-run the setup_keptn.sh
script.
2) The Git PAT token doesn't have full repo access
3) This is a temporary / transient issue with the connection to GitHub (see below for explanation)
When the setup_keptn.sh
creates the project, it issues the keptn create project
command via CLI. The logic here might need improvement in case of failure. I'm not sure, it may already retry on failure but the repo definitely does exist (it is created independently by the gh
CLI command) so something is clearly going wrong.
From there, all the other errors are cascading symptoms. If the Keptn project doesn't exist, everything else is doomed to fail.
Hi @agardnerIT
Regarding Slow startup time - I suggest adding a line into the script to tell the user "The job-executor-service takes up to 10 minutes to start; please be patient"
On the create project issue - item 1 is not it - I tried twice to make sure there was not a preexisting repo. Item 2 - is there some other access I need to enable besides "repo"?
I tried again after adding almost every conceivable permission to the token. Still get same failure. A new GitHub repo was created but it is empty.
Starting to create project
Error: Create project was unsuccessful.
upstream repository not found
Starting to create service
Create service was unsuccessful
Error: Create service was unsuccessful. could not get stages of project fulltour: project not found
Could it perhaps be that my username on GitHub is different than the userid? Grasping at straws.
ubuntu $ git clone https://github.com/LeslieMurphy/keptdemo3.git Cloning into 'keptdemo3'... Username for 'https://github.com': les.murphy@xxxxxxxxxxxxxxx.com Password for 'https://les.murphy@xxxxxxxxxxxxxxx.com@github.com': remote: Repository not found. fatal: repository 'https://github.com/LeslieMurphy/keptdemo3.git/' not found
ubuntu $ git clone https://github.com/LeslieMurphy/keptndemo3.git Cloning into 'keptndemo3'... warning: You appear to have cloned an empty repository.
Oh... It could? Tbh I didn't know that was possible with GitHub :)
I will try again and see if it accepts my username with the token
set_git_details.sh
currently creates the URL as:
export GIT_REPO=https://github.com/$GIT_USER/$GIT_NEW_REPO_NAME.git
So you might well have solved it.
Also, there was a typo above. You wrote keptdemo3
and it is keptndemo3
with an n
I know! I fat fingered it! It is WORKING when I use my username!
GitHub allows authentication using either the Username or email address. If the email is used in the Keptn script for the token credentials, the script will create a repo but subsequently will fail. Is there a way to improve the process to make it work in either case?
Here is a summary of the issue:
@LeslieMurphy we are updating to Keptn v1.1.0 and it might be nice to include this in that work (#74).
If I'm not mistaken, all we need here is to modify the set_git_details.sh
and include a new additional parameter for GIT_EMAIL. Then if it is NOT set, we assume username and just use that, but if it IS set, we can accomodate your usecase.
Make sense?
Hi Adam, are you proposing three prompts - one for email and one for username plus the token? Could you instead extract the username from github after creating the repo?
If you first install jq this will do the trick:
gh api 'https://api.github.com/user' | jq .login
Result after using my github email address for authentication is:
"LeslieMurphy"
That's a much better idea. Thanks!
Would you like to raise the PR (I can help) or should I do it?
@LeslieMurphy anything more to do here or can we close?
I think we are all good on it. Sent from my iPhoneOn Mar 26, 2023, at 5:25 PM, Adam Gardner @.***> wrote: @LeslieMurphy anything more to do here or can we close?
—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: @.***>
There have now been 2 reports of the Keptn demo being broken: