Bump all dependencies to latest to upgrade controller-runtime #1050
Conversation
| @@ -87,9 +86,10 @@ func main() { | |||
|
|
|||
| ctrl.SetLogger(zap.New(zap.UseFlagOptions(&opts))) | |||
| ctrlOpts := ctrl.Options{ | |||
| Scheme: scheme, | |||
| MetricsBindAddress: metricsAddr, | |||
| Port: 9443, | |||
There was a problem hiding this comment.
This is the default of the Webhook server created in the new version as well.
|
@markusthoemmes Thanks again for the PR. |
This upgrades all dependencies of the project to their latest releases. The most notable of those is controller-runtime, which has seen some significant rework. No semantic changes are intended in this change. Signed-off-by: Markus Thömmes <[email protected]>
markusthoemmes
force-pushed
the
upgrades
branch
from
4d38efa to
501e36c
Compare
|
Done. I was staging PRs and didn't notice the other one actually merged. |
| ctrlOpts.Cache.DefaultNamespaces = make(map[string]cache.Config, len(namespaces)) | ||
| for _, ns := range namespaces { | ||
| ctrlOpts.Cache.DefaultNamespaces[ns] = cache.Config{} |
There was a problem hiding this comment.
|
Hmm. Gotta fix the code generation. Will take care of that. |
|
might need |
Signed-off-by: Markus Thömmes <[email protected]>
Signed-off-by: Markus Thömmes <[email protected]>
|
Hmm. This diff is becoming much larger than intended because the PodSpec grew quiet a bit between versions, I suppose. |
Signed-off-by: Markus Thömmes <[email protected]>
| if [[ $ret -eq 0 ]] | ||
| then | ||
| echo "${DIFFROOT} up to date." | ||
| else | ||
| echo "${DIFFROOT} is out of date. Please rerun make generate" | ||
| echo "${DIFFROOT} is out of date. Please run hack/update-codegen.sh" |
There was a problem hiding this comment.
This had to be changed because generate-groups.sh seems to delete the tree now and thus the comparison was wrong.
This change essentially copies the current tree into a temp dir, then generates the clients "as normal" (full on replacing the tree) and then diffs the copied out tree with the now regenerated clients. Has the added benefit, that the actual generation is deduplicated.
|
@benjaminhuo As mentioned above, this diff has grown quite a bit. We could try to make it smaller by upgrading controller-runtime in smaller steps or trying to break the K8s upgrades out. Let me know if you'd like me to invest some effort into trying to make the diff smaller. |
@wanjunlei what do you think? |
I think it's more appropriate to upgrade the controller runtime in smaller steps. |
We might need to catch up with the upstream k8s community updates and drop supports for lower version K8s. |
we really need to catch up with upstream now because there're quite big differences and the code cannot compile with latest k8s dependencies:
|
|
Fixed by #1251, close this one |
What this PR does / why we need it:
This upgrades all dependencies of the project to their latest releases. The most notable of those is controller-runtime, which has seen some significant rework. No semantic changes are intended in this change.
Does this PR introduced a user-facing change?
Additional documentation, usage docs, etc.: