ShimV2 has shim.Delete command to cleanup task's temporary resource,
like bundle folder. Since the shim server exits and no persistent store
is for task's exit code, the result of shim.Delete is always 137 exit
code, like the task has been killed.
And the result of shim.Delete can be used as task event only when the
shim server is killed somehow after container is running. Therefore,
dockerd, which watches task exit event to update status of container,
can report correct status.
Back to the issue #6429, the container is not running because the
entrypoint is not found. Based on this design, we should not send
137 exitcode event to subscriber.
This commit is aimed to remove shim instance first and then the
`cleanupAfterDeadShim` should not send event.
Similar Issue: #4769Fix#6429
Signed-off-by: Wei Fu <fuweid89@gmail.com>
The build number used to determine whether we need to pull the Windows
Server 2022 image for the integration tests was previously hardcoded as there
wasn't an hcsshim release with the build number. Now that there is and it's
vendored in, this change just swaps to it.
Signed-off-by: Daniel Canter <dcanter@microsoft.com>
This tag brings in some bug fixes related to waiting for containers to terminate and
trying to kill an already terminated process, as well as tty support (exec -it) for
Windows Host Process Containers.
Signed-off-by: Daniel Canter <dcanter@microsoft.com>
This is a effective revert of 294143bf38
The one thing that makes it not a total revert is this keeps the usage
of chain interceptors, which prevents us from overwriting interceptors
passed into client options.
The automatic trace injection is unnecessary overhead since callers of
this function can add the necessary interceptors when creating the
client.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
The global &daemon{} object only gets an address assigned if you call
start() on it, which only gets called if you don't pass no-daemon. The
aftermath of this is that running the integration tests with no-daemon
just has them fail trying to create a client for containerd.
This change just assigns whatever address is passed to the binary even in
the no-daemon case so you can run the integration tests against an already
running containerd instance.
Signed-off-by: Daniel Canter <dcanter@microsoft.com>
OCI hooks aren't implemented on Windows. The test will, and has been,
actuallyrunning fine on Windows because the Github runners seem to have
a 'ps' binary in the users PATH, but there's not any actual hook
functionality being tested as any of the OCI fields are ignored for
Windows containers.
Signed-off-by: Daniel Canter <dcanter@microsoft.com>
Skip this test until this error can be evaluated and the appropriate
test fix or environment configuration can be determined.
Signed-off-by: Derek McGowan <derek@mcg.dev>
The OCI image spec did a v1.0.2 security release for CVE-2021-41190, however
commit 09c9270fee, depends on MediaTypes that
have not yet been released by the OCI image-spec, so using current "main" instead.
full diff: 5ad6f50d62...693428a734
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
GA for ws2022 github actions VMs launched a couple weeks ago so seems like
it's time to try out the CI on this new SKU.
This involved adding new ws2022 runs for the OS matrices in the CI, fixing up
a test in the platforms package and adding a mapping for the ws2022 container image in
integration/client.
Signed-off-by: Daniel Canter <dcanter@microsoft.com>
With the ghcr images now built and working, switch over to
use these new images and update the default name.
Signed-off-by: Derek McGowan <derek@mcg.dev>
This change skips the TestExportAndImportMultiLayer in integration/client
for the time being. It seems the image was updated recently and no longer
has a Windows entry in the manifest so the test will always fail. This should
be reverted when we figure out what happened to the image, but this is to
unblock PRs for the time being.
Signed-off-by: Daniel Canter <dcanter@microsoft.com>
This tag contains some changes for the Windows shim for retrying
stdio named pipe connections if containerd restarts. It also is built with v1.1.0 of
ttrpc which has some fixes for a deadlock we'd observed on Windows.
Signed-off-by: Daniel Canter <dcanter@microsoft.com>
This tag contains a fix for a deadlock observed when there are multiple
simultaneous requests from the same client connection.
Signed-off-by: Daniel Canter <dcanter@microsoft.com>
Use the time for the last non-running status to determine
whether the restart did not occur as expected. The
current timestamp only accounts for when the running
status was seen, however, the restart would have always
occurred in between the previous check and latest check.
Therefore, it makes more sense to use the previous check
to determine whether a failure was seen from the restart
monitor not restarting as expected.
Signed-off-by: Derek McGowan <derek@mcg.dev>
* Bump k8s.io/cri-api to latest version - v0.23.0-alpha.4
* Vendor github.com/vishvananda/netlink for network stats
Signed-off-by: David Porter <porterdavid@google.com>
restart monitor test was failing due to occasionally taking past the
deadline on windows tests. Add a small additional grace period to
deflake the test.
Signed-off-by: David Porter <porterdavid@google.com>
