oc cluster up fails on latest master #16995
Assignees
Labels
Comments
|
@openshift/sig-developer-experience |
|
@openshift/sig-master related to refactoring of admission plugins? |
|
use upstream ,not found the bug. are you sure? |
|
@xiaods I was able to recreate it. Presumably this means the master config file being generated by the latest origin code is not valid for a 3.7-alpha1 master/cluster? Oddly it works fine with a 3.6 cluster, so did we just have a temporary regression in 3.7-alpha1? (in which case presumably we can close this since we shouldn't be making any compatibility assumptions about an alpha release). |
|
@deads2k ptal |
|
I'm betting its an old images problem. Admission webhooks wiring was messed up and I fixed it in master. |
|
But it worked before, so it has to be config related (some new config we
added / took away).
…On Mon, Oct 23, 2017 at 3:19 PM, David Eads ***@***.***> wrote:
I'm betting its an old images problem. Admission webhooks wiring was
messed up and I fixed it in master.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#16995 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABG_p2Oegr_Jkz9gPWP4lisvkO6sYA-5ks5svJJPgaJpZM4QCKUq>
.
|
|
This #16861 enabled admission webhooks in |
|
Is there any mileage in improving the process around releasing openshift/origin:vx.y.z-alpha.* images so that this kind of thing can be avoided / can be fixed quicker when it happens? |
|
Unless you're volunteering, not right now. It's mostly blocked on the
other more critical CI automation.
…On Mon, Oct 23, 2017 at 5:08 PM, Jim Minter ***@***.***> wrote:
Is there any mileage in improving the process around releasing
openshift/origin:vx.y.z-alpha.* images so that this kind of thing can be
avoided / can be fixed quicker when it happens?
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#16995 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABG_pxm4Td3JnDVZnxEtmqMHk0w1utq1ks5svKwAgaJpZM4QCKUq>
.
|
pweil-
added
component/cluster-up
kind/bug
|
also failing against OSE v3.3 with a different error: |
|
also failing on ose v3.5 with a different error: |
|
and in what is probably a surprise to no one, v3.4 is also broken: |
|
(v3.6 is currently broken too, but i'm fixing that here: https://github.com/openshift/origin/pull/17029/files#diff-3e0aea9639b40036cf195d2ad1a7141dR713) |
|
@bparees it was my understanding that we weren't going to necessarily maintain backwards compatibility with previous versions (if you want to run those, you use the older oc client). If that's the case, should we worry about the current client not being able to start older versions? |
we never really got a sign off on that, so as of today there is still an expectation that this works. At a minimum we probably have to enforce an N-1 rule like we do for using an oc client to talk to a cluster, but we don't even have sign off on that and as soon as the CDK team starts trying to test for 3.7 they're going to discover this and open bugs. |
|
^^ @deads2k |
|
@deads2k if you know what's incompatible can you point me to it? |
@csrwng comment #16995 (comment) suggests that @bparees is on it. The other one has to do with the server flag enabling |
and yeah, we agree and are trying to get consensus from downstream consumers of oc cluster up about this, but we aren't there yet. |
|
Given that the latest oc client works with latest, 3.7.0-rc.0 and 3.6.0, I am closing this issue as resolved. The current oc client itself is not compatible with versions older than 3.6 because of the switch to API groups. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The text was updated successfully, but these errors were encountered: