This is not removing any tests, we moved the tests to a different folder
and had to keep this folder around for CI to pass. This is just a
cleanup

…ecks (#623)

Signed-off-by: Chun Lin Yang <clyang@cn.ibm.com>

Signed-off-by: Chun Lin Yang <clyang@cn.ibm.com>

* have a test

* Revert "have a test"

This reverts commit bd6d803a9ed81869e27ebff96f9c9d66678b9c56.

* improved compatability of unit test "TestEnvoyStatsCompleteAndSuccessful"

* addressed an issue of namespace configuration in CI tests

* improved robust of integration test

* fixed compatibility issue in 'mirror_test.go'

* improved compatibility of CI tests in 'util.go'

* Revert "improved compatibility of CI tests in 'util.go'"

This reverts commit ae9b612b2416d640e58afb4850d109eb11e5e1f9.

* fixed bugs in building with docker image

* format Makefile

* remove docker run terminal emulator because there is no need

* remove unnecessary env var

* fixed format issue

* removed 'localhostAddr' configuration in probe because empty host will be added in 'stats.go' automatically

* add host if it is empty

* removed un-used tests compared to the community mater

* adopt to gobin

* fixed corrupted docker building process

* format fixes

* fixed format

* remove changes in docker executing command

* remove commands in docker executing command

* rollback make file changes

Fixes prometheus/prometheus#6512

This does NOT skip the test -- the test was already skipped, but was
done at the incorrect spot such that the test setup still occurs

* Move proxy version match against envoyfilter to top level

* fix ut

* fix ut

* remove unused proxy

* fix ut

* fix benchmark

* fix lint

* fix lint

* remove the return

…722)

This patch removes `mixer/adapter/kubernetesenv/config/.yaml` which is
probably created by mistake. The actual file should be
`mixer/adapter/kubernetesenv/config/kubernetesenv.yaml` and already exist.

The crash was introduced by #19786

Fixes #19806

* Fix args.wait not being used.

* Fix import.

* Add all securityContext fields in injected containers

Fixes #17318

* Update injection unit tests

Fixes #17318

* fix TestController_GetPodLocality flake

* run sub test in parrallel

* close channel

* typo

* minor ads improvements

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* revert IsApplicationNodetype

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* lint

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* address review comments

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* lint

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* remove unnecessary code

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* lint

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* Add testing for EndpointSlice

* Fix mode

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* Add TODO comments of version 1.5 for translate

* make fmt.

This reverts commit a42d9e7.

Fixes #19834

Should fix #19846

…or naming consistent in e2e testing framework. (#19768)

Currently we generate invalid JSON, since we have no comma between the
list. The easiest way to do this is to have a toJson function so we
don't need to hack together json marshalling in the template. There are
no compatibility issues because the bootstrap logic and bootstrap file
are tightly coupled

Currently SDS logging is inconsistant with Pilot/pilot-agent. This was
not as noticable before, but now that they are in the same deployment
its very noticeable and makes parsing the logs harder.

This change is a step towards consistency with the rest of Istio and
reduces the verbosity substantially.

Fixes #20279

… defined as tls or https (#20248)

* move trafficDirection to the listener options

* disable timeout for tls and https port

* Revert "move trafficDirection to the listener options"

This reverts commit 6a1aec9.

* fix

* update comment

* ..

…JWT (#20179)

- Support validating first party JWT for platforms without third party JWT.
- Add the JWT policy configuration to Operator.
- Config grpc calling credentials.

* Remove certmanager integration

This has been deprecated and removed from all docs for the past 2
releases. It is untested and does not work anymore (due to deprecation
of this version of certmanager).

* Regen golden

* Disable policy deployment by default

* Update goldens

* Fix tests

* Disable policy checks fix

* Fix default e2e values

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* add config validation to pilot

* update manifests

* fix clusterrole permissions

* update operator test data

* Update tests/integration/galley/webhook/webhook_test.go

Co-Authored-By: John Howard <howardjohn@google.com>

* update test framework to wait on pilot instead of galley

* more test fixes

* use pilot webhook port

* fix cert load

* undo test changes

* remove unused var

* skip key/cert reload test. istiod doesn't support rotation.

* skip istioctl webhook test until its updated to work with istiod

* update webhook tests to support istiod and non-istiod cases

* fix istiod check

* check override values

* gencheck

Co-authored-by: John Howard <howardjohn@google.com>

Today, citadel will never shutdown on SIGTERM, only SIGKILL. With
default settings, this means 30s to shutdown (termination grace period =
30s by default). This also impacts tests.

The root cause of this is that we have a couple things blocking
shutdown. First, we do have some attempt at catching the SIGTERM, but it
only actually applies after ~45min of time.Sleeping(). Instead, we
should allow the stop channel to interupt that sleep.

The other problem is that when the monitoring server closes, it will
write an err on an unbuffered channel. In the normal case where things
exit graceful, there is a race between exiting because the stop channel
is closed, and actually reading from the monitoring error channel. To
get around this, we make the channel a buffered channel so we can write
asyncronously. This way if there is a real error, we will still see it,
but on shutdown we don't block.

* Stopgap for Galley dashboards with istiod

Right now on Istiod the galley dashboard is not useful because there is
no galley. In the 1.5 timeframe it may not be feasible to refactor our
dashboards to be feature oriented rather than component (if we even
desire this). In the short term, we can make all of the metrics shared
between istiod and galley not have a job= label. This will allow it to
pick up either source of the metric. We cannot do this for the other
metrics, as they are generic like CPU, etc.

* gen

* ensure ads refuses connections when caches are not synced

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* move log lines to go func

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* istio bug fixes

* added some comments

* fixed to pass the integration test

* go fmt

Regression since #20213

We turned on policy, but disabled checks. Didn't notice since it only
runs in postsubmit.

Regression since #20213

We turned on policy, but disabled checks. Didn't notice since it only
runs in postsubmit.

* istioctl describe: fixes, test case, remove deprecated code

* Remove unneeded subtest structure

* Remove unnecessary change in describe

* Match describe output by regexp, not string

* Additional output to test the tests

* Keep constructor signature

* Fix compilation problem

* Look for Envoy config metadata in Istio 1.5.x format

* 'make format'

* Ran 'gen'

* istioctl describe: fixes, test case, remove deprecated code

* Remove unneeded subtest structure

* Remove unnecessary change in describe

* Match describe output by regexp, not string

* Additional output to test the tests

* Keep constructor signature

* Fix compilation problem

* Look for Envoy config metadata in Istio 1.5.x format

* 'make format'

* Ran 'gen'

* istioctl describe: fixes, test case, remove deprecated code

* Remove unneeded subtest structure

* Remove unnecessary change in describe

* Match describe output by regexp, not string

* Additional output to test the tests

* Keep constructor signature

* Fix compilation problem

* Look for Envoy config metadata in Istio 1.5.x format

* 'make format'

* Ran 'gen'

* Set Citadel as the default provider for Pilot certificate

- As some platforms may not have k8s signing APIs,
set Citadel as the default provider for Pilot certificate.
- Integration tests when pilotCertProvider=citadel

* Fix the webhook config controller with the citadel provider

* Fix the file permission problem

Fix the file permission problem:
error	failed to create discovery service: grpcDNS: open ./var/run/secrets/self-signed-root.pem: permission denied

* Write the root cert to a directory

* use correct caBundle for mutating webhook

Co-authored-by: Jason Young <jasonyoung@google.com>

Right now if you have a lot of endpoints Pilot logs are extremely
spammy. We log every IP every time a new IP is added, which for high #
of endpoints is often 1000s and very often. Then we log 1000s of lines
about EDS incremental update. I think those belong in debug to make the
logs a bit more sane.

* fix unused operator init wait option.

* fix test.

…20327)

* refresh push context after cache sync

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* pass env

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* call push after cache sync

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* - Move health check to single config directory to avoid multiple setup
- Add separate test for global, annotation and separate port

* Verify test output

* Remove sleep and run each test in separate namespace

* Remove duplicate namespace

* Revert config map for http probe

* handle http proxy protocol in outbound listeners

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

* add comments and return

Signed-off-by: Rama Chavali <rama.rao@salesforce.com>

Clean up a bunch of top level junk, like common files stuff, license (we
use the top level license, nothing is changing legally here), legacy
tests we no longer run, and kustomize. This was all stuff left over from
the repo merge that is unused.

…ion (#20339)