This one-line fix is already in main, from #31862. but that PR relies on other refactor PRs, and
all are too broad to apply to the patch release branch.
Signed-off-by: Scott Rigby <scott@r6by.com>
`helm upgrade --dependency-update` built its downloader.Manager without a
RegistryClient, unlike the identical literals in install, dependency update,
and dependency build. Resolving an OCI chart dependency whose version is a
range then dereferenced a nil *registry.Client in registry.(*Client).Tags,
panicking the process. Pass the registry client already constructed for the
command, matching install.
Add a regression test that runs `upgrade --dependency-update` against a chart
with an unresolved OCI dependency and asserts the command returns an error
instead of panicking.
Signed-off-by: Gates Wang <9372086+SetagGnaw@users.noreply.github.com>
(cherry picked from commit 0604d8fb1e)
Co-authored-by: George Jenkins <gvjenkins@gmail.com>
Signed-off-by: Will Noble <4642422+ouillie@users.noreply.github.com>
(cherry picked from commit 61bba7c659)
* Fix missing conflict retry with server-side apply
Signed-off-by: Jakub Jaruszewski <jjaruszewski@man.poznan.pl>
* Only retry on resourcequota transient conflict
Signed-off-by: Jakub Jaruszewski <jjaruszewski@man.poznan.pl>
* Use isServerSideRetryable + naming nitpick
Signed-off-by: Jakub Jaruszewski <jjaruszewski@man.poznan.pl>
---------
Signed-off-by: Jakub Jaruszewski <jjaruszewski@man.poznan.pl>
(cherry picked from commit aa1ae3a360)
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Signed-off-by: kimsungmin1 <142377392+kimsungmin1@users.noreply.github.com>
(cherry picked from commit d539556a8d)
- Use (*net.ListenConfig).Listen with the test context instead of
net.Listen (noctx).
- Use suite.NoError instead of suite.Nil for the error check
(testifylint).
- Drop the unnecessary leading blank lines in the scope test funcs
(whitespace).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Signed-off-by: kimsm28 <sm28.kim@samsung.com>
(cherry picked from commit 503acff975)
The scope test configured a TLS registry while advertising an http token
realm. oras refuses to send credentials to an http token realm when the
registry itself was contacted over https, so the auth server was never
reached and the test timed out ("timeout waiting for auth request").
Use a plain-http registry so the registry and token-realm schemes match;
the scope carried in the token request is what this test verifies, and
that is independent of TLS.
Also make the auth handler's channel send non-blocking so that a client
retry can never block the handler and stall the push/pull flow, which
addresses the review comment about a potential deadlock.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Signed-off-by: kimsm28 <sm28.kim@samsung.com>
(cherry picked from commit 3cbee7e935)
- Fix data race by using channel instead of shared variable for requestURL
- Make assertions more robust by parsing URL and checking query parameters instead of exact string matching
Signed-off-by: kimsm28 <sm28.kim@samsung.com>
(cherry picked from commit 9b7a70f0ca)
- Change 'bellow' to 'below' (spelling correction in multiple places)
- Change 'WithScopeHint' to 'withScopeHint' to match actual function name
Signed-off-by: kimsm28 <sm28.kim@samsung.com>
(cherry picked from commit ee181f4570)
- Change DockerRegistryHost to use 127.0.0.1 for HTTP tests and helm-test-registry for TLS tests to match certificate hostname
- Add defer ln.Close() to prevent resource leak in token auth server setup
- Restore requestURL variable and assertion in test body instead of handler to avoid potential race condition
Signed-off-by: kimsm28 <sm28.kim@samsung.com>
(cherry picked from commit 1f0a728496)
- only cache kube client when creation succeeds
- add regression test for two failed reachability checks
Signed-off-by: Mohammad Abdolirad <m.abdolirad@gmail.com>
(cherry picked from commit 1f7869c31f)
Return the actual error from io.Copy in Digest() instead of nil.
The previous code swallowed the error and returned an empty string
as a valid SHA-256 digest, which could silently break chart
provenance verification.
Also fix encodeRelease() in pkg/storage/driver/util.go:
- Close the gzip writer on the w.Write() error path to avoid
leaking resources.
- Check the error return from gzip.Writer.Close(), which flushes
remaining compressed data and can fail.
Assisted-by: Grok/xAI
Signed-off-by: Sebastien Tardif <sebtardif@ncf.ca>
(cherry picked from commit e095e690a8)
- Close log stream after reading (prevents connection/fd leak)
- Strengthen tests to assert on output headers rather than error paths
- Remove unused import
Signed-off-by: Sebastien Tardif <SebTardif@ncf.ca>
(cherry picked from commit 922558fc1a)
When a test pod contains multiple containers (e.g. Istio/Consul/Vault
sidecars), 'helm test --logs' failed with 'a container name must be
specified'. This happened because GetPodLogs called the Kubernetes log
API without specifying a container name.
The fix fetches the pod spec first, then iterates over all containers
(init containers + regular containers) and requests logs for each one
explicitly. Errors from individual containers are collected and returned
together via errors.Join rather than aborting on the first failure.
Also fixes a typo: hooksByWight -> hooksByWeight.
Closes#6902
Signed-off-by: Sebastien Tardif <SebTardif@ncf.ca>
(cherry picked from commit 854f7f6b72)
* fix: protect FailingKubeClient.RecordedWaitOptions from concurrent access
Add a sync.Mutex to guard the append to RecordedWaitOptions in
GetWaiterWithOptions, fixing a data race detected by -race when
concurrent goroutines (e.g. upgrade + rollback) both call
GetWaiterWithOptions on the same FailingKubeClient instance.
Fixes race failures in TestUpgradeRelease_Interrupted_RollbackOnFailure
and TestInstallRelease_RollbackOnFailure_Interrupted.
Signed-off-by: Terry Howe <thowe@nvidia.com>
* fix: extract appendRecordedWaitOptionsLocked helper with defer unlock
Signed-off-by: Terry Howe <terrylhowe@gmail.com>
---------
Signed-off-by: Terry Howe <thowe@nvidia.com>
Signed-off-by: Terry Howe <terrylhowe@gmail.com>
(cherry picked from commit a5552edf9f)
Commands like 'helm registry login', 'helm push', and 'helm pull' were
writing success messages ("Login Succeeded", "Pushed:", "Pulled:",
"Digest:") to stderr instead of stdout. The root cause was that
newDefaultRegistryClient and newRegistryClientWithTLS hard-coded
os.Stderr as the registry client writer, ignoring the out io.Writer
that main() passes as os.Stdout.
Thread out io.Writer through newRegistryClient, newDefaultRegistryClient,
and newRegistryClientWithTLS, and update all call sites in pkg/cmd.
Fixes#13464
Signed-off-by: Terry Howe <terrylhowe@gmail.com>
(cherry picked from commit c2f1b238a1)
`isVersionRange` checked for `x`/`X` across the entire version
string, misclassifying exact versions like `1.0.0-fix`,
`2.0.0-next`, or `1.0.0+exp` as ranges.
Signed-off-by: Benoit Tigeot <benoit.tigeot@lifen.fr>
(cherry picked from commit 740174a2b1)
From Matt's comment
> The check for " || " should remove the spaces and have "||". Spaces around the || aren't required.
Signed-off-by: Benoit Tigeot <benoit.tigeot@lifen.fr>
(cherry picked from commit b79d7f1881)
When using version ranges like ^1 or ~1.10, Helm incorrectly showed
warnings about falling back to closest version. Only show the warning
when an exact version is requested but not found.
Fixes: https://github.com/helm/helm/issues/31757
Signed-off-by: Benoit Tigeot <benoit.tigeot@lifen.fr>
(cherry picked from commit 1e145ee2b2)
The previous change suppressed ctx.Err() whenever there were no
resource-specific errors, which incorrectly swallowed context.Canceled
and other non-deadline errors signalling an external interruption.
Refine the condition: only suppress context.DeadlineExceeded when there
are no resource-specific errors (resources are Unknown/NotFound, meaning
the delete wait succeeded or resources were already gone). Any other
context error — including context.Canceled — is always propagated.
Signed-off-by: Terry Howe <terrylhowe@gmail.com>
(cherry picked from commit 5e09ee78ee)
During informer initialization there is a brief window where watched
resources appear as Unknown before their real statuses are delivered.
The statusObserver skips Unknown resources when waiting for deletion
(they may have been deleted before the watch started), but if *all*
resources are in that transient Unknown state the skipped-resource list
is empty. AggregateStatus on an empty slice returns the desired status,
causing cancel() to be called immediately — before any real status event
has arrived.
Guard against this by tracking the count of Unknown-skipped resources.
When every resource was Unknown-skipped and none have a definitive status
yet, defer the early-cancel decision until at least one resource reports
a real status. This preserves the correct behaviour for resources that
were genuinely deleted before the watch started (they eventually receive
a NotFound or stay Unknown, and the aggregate succeeds), while fixing
the race for resources that are transiently Unknown at startup.
Also tighten the ctx.Err() check in waitForDelete: only append a
deadline error when there are resource-specific errors to accompany it.
A timeout while all resources are Unknown or NotFound is not itself an
error — the resources are in an acceptable state for a delete wait.
Fixes: TestStatusWaitForDelete/error_when_not_all_objects_are_deleted
Signed-off-by: Terry Howe <terrylhowe@gmail.com>
(cherry picked from commit 4e24ee41a4)
Scott Rigby
Gates Wang
Will Noble
Will Noble
Jakub Jaruszewski
kimsungmin1
kimsm28
Terry Howe
Mohammad Abdolirad
Sebastien Tardif
Jeaeun Kim
Matheus Pimenta
George Jenkins
Benoit Tigeot