The test relied on a 100ms sleep to ensure that controller was done. If that
race was lost, one goroutine was intentionally prevented from completing by
locking a mutex permanently. A TODO was left about detecting that.
Adding goroutine leak checking in
https://github.com/kubernetes/kubernetes/pull/126387 revealed that this race
indeed sometimes is lost because the goroutine
leaked (https://github.com/kubernetes/kubernetes/issues/129400).
Waiting for controller shutdown instead of relying on timing should fix this.
* accept GINKGO_FLAGS for test-e2e-node.sh
* Update Makefile to document LABEL_FILTER
* --label-filter part of ginkgoflags
* Update Makefile
* if label-filter then avoid skip and focus defaults
* LABEL_FILTER can be empty
* Update build/root/Makefile
Co-authored-by: Patrick Ohly <patrick.ohly@intel.com>
* skip defaults only if label-filter not set
Co-authored-by: Patrick Ohly <patrick.ohly@intel.com>
* focus and label_fiter can live together
Co-authored-by: Patrick Ohly <patrick.ohly@intel.com>
* skip and label_filter can live together
Co-authored-by: Patrick Ohly <patrick.ohly@intel.com>
---------
Co-authored-by: Patrick Ohly <patrick.ohly@intel.com>
NewTypedDelayingQueueWithConfig spawns a goroutine, but apparently shutdown is
already handled somehow. Therefore only the option to set a logger gets added
to the Config struct.
The problem then becomes that developers might forget to set that
logger. logcheck can't detect that. For now, all in-tree users get updated
immediately.
CreateOrRetain is supposed to operate on an object name which isn't
necessarily the given object's name (for use in migrations), this
restores that feature.
Replace all uses of deprecated functions with their generic variants.
Providing the context externally isn't useful right now, drop it from
the new functions and use context.Background() where needed.
Signed-off-by: Stephen Kitt <skitt@redhat.com>
This uses generics to generalise the various CreateOrUpdate,
CreateOrRetain etc. functions. Where appropriate, the context is added
as an initial argument to the new functions.
ConfigMapMutator isn't used anywhere else, so it's dropped in favour
of the private objectMutator added in this commit.
Signed-off-by: Stephen Kitt <skitt@redhat.com>
