Closed johnsca closed 6 years ago
I deployed this and it got to after-deploy
and failed with:
Failure in step 01_install_arti after-deploy
But I don't see anything in the ~/.cache/conjure-up/canonical-kubernetes/addons/jfrog/steps/01_install_arti/after-deploy.out
or ~/.cache/conjure-up/canonical-kubernetes/addons/jfrog/steps/01_install_arti/after-deploy.err
files that looks like reasons for it to have failed.
It looks like the channel was set properly on both charms, though:
[johnsca@murdoch:~] $ juju config kubernetes-master channel
1.9/stable/jfrog
[johnsca@murdoch:~] $ juju config kubernetes-worker channel
1.9/stable/jfrog
So I think this PR is good, as long as the change is ok with @shrajfr12
I would like to know why it thinks it failed, though.
Hi @johnsca, what are the contents of ~/.cache/conjure-up/canonical-kubernetes/addons/jfrog/steps/01_install_arti/jfrog_installer.log ? Most likely it is a tiller not found or available issue. If that is the case, just rerunning the helm command in the after-deply.err should fix it. This is a helm-tiller issue which happens every now and then
But I will also try to with the these changes and let you know
I ran in my local environment without passing bundle-add channels.yaml since it is now in before-config (I had to chmod +x before-config as I manually created the file and it did not hv execute permissions) and install went through fine and verified from kubectl get nodes and confirmed it shows 1.9.6. So looks good and thanks for updating README as well to reflect this change.
Yes, that was the error, "could not find a ready tiller pod." So it sounds like you're ok with this approach, so I'll go ahead and merge and update edge.
Is there anything I can do to help address the occasional missing tiller issue?
This is now available in edge (rev 998)
Needs review from @shrajfr12 and I haven't tested it yet (starting it now, will update)