Kubeadm: kinder: stop calling etcdctl from within the etcd containers
with all control-plane images and etcd going for distroless, see:
kubernetes/kubernetes#70249
we need to stop executing etcdctl from within the etcd containers in kinder.
potential solution is to get the etcdctl version that we want from the etcd tarballs and execute it from a temporary directory:
https://github.com/etcd-io/etcd/releases
cc @fabriziopandini
neolit123
All 15 comments
Another option is described here https://mauilion.dev/posts/etcdclient/
fabriziopandini
on 10 Nov 2019
Another option is described here https://mauilion.dev/posts/etcdclient/
this would require us to build our own image that has etcdctl,
the k8s.gcr.io/etcd:VERSION image will not have it at some point in the future.
neolit123
on 10 Nov 2019
pro and cons.
considering the latest breaking changes in etcd, I would prefer to not rely on etcd tarballs...
fabriziopandini
on 11 Nov 2019
the breaking changes manifested without relying on tarballs.
if we start pulling tarbals to execute etcdctl we still need the different CA flags for etcdctl.
same goes if we start maintaining an etcdctl image - it still needs different CA flags.
i'd prefer the tarballs, but ideally the etcd and etcdctl binaries should be hosted in a separate tarballs.
neolit123
on 21 Nov 2019
Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.
If this issue is safe to close now please do so with /close.
Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale
fejta-bot
on 19 Feb 2020
/remove-lifecycle stale
neolit123
on 19 Feb 2020
Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.
If this issue is safe to close now please do so with /close.
Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale
fejta-bot
on 19 May 2020
/remove-lifecycle stale
looking at the WIP PR for distorless etcd image, it seems that the etcdctl is preserved and etcd is not the only binary. will continue tracking this issue until the work there is complete.
neolit123
on 19 May 2020
Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.
If this issue is safe to close now please do so with /close.
Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale
fejta-bot
on 17 Aug 2020
/remove-lifecycle stale
fabriziopandini
on 17 Aug 2020
It seem etcdctl is going to stay in the image, since others depend on it
too. but we can keep this open for longer if they change their mind.
neolit123
on 17 Aug 2020
Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.
If this issue is safe to close now please do so with /close.
Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale
fejta-bot
on 15 Nov 2020
/remove-lifecycle stale
fabriziopandini
on 15 Nov 2020
Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.
If this issue is safe to close now please do so with /close.
Send feedback to sig-contributor-experience at kubernetes/community.
/lifecycle stale
fejta-bot
on 13 Feb 2021
/remove-lifecycle stale
neolit123
on 13 Feb 2021
Related issues
jbrandes
路
4Comments
bruceauyeung
路
4Comments
mattmoyer
路
4Comments
kvaps
路
3Comments
chuckha
路
3Comments
Most helpful comment
/remove-lifecycle stale
looking at the WIP PR for distorless etcd image, it seems that the etcdctl is preserved and etcd is not the only binary. will continue tracking this issue until the work there is complete.