Seems that at least some versions of container, when using ConifgPath,
still rely on the runtime options and its APIs from the not in use
anymore github.com/containerd/cri-containerd/pkg/api/runtimeoptions/v1.
The fact backward compat breaks when moving from the old to the new
runtime options, which happened as part of f60641a6e6d, strongly feels
like a containerd bug. Regardless, we can easily work this around on
our side without much hassle.
Just by importing old runtime options the unmarshalling doesn't break
anymore and we can easily check whether getting the options fails or not
and fallback to the old way if it does.
Fixes: #2258
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
Commit 32c9ae1388 upgrade the
containerd vendor, which used the socket path to replace
the abstract socket address for socket listen and dial, and
there's an bug in containerd's abstract socket dialing.
Thus we should replace our monitor and exec socket server
with the socket path to fix this issue.
Fixes: #2238
Signed-off-by: fupan.lfp <fupan.lfp@antgroup.com>
Let's add this target so we can actually enforce, as part of the static
checks (which will be added in a follow-up commit), that our vendored go
code is up-to-date.
Related: #2159
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
This will help us to ensure that we always update the vendored code when
needed. Right now we've been lacking behind and we tend to realise
something change during the next mandatory update, which is not exactly
optimal.
Related: #2159
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
Use a defined error variable replade inplace error, and shortcut
for handling errors returned from function calls.
Fixes: #2187
Signed-off-by: bin <bin@hyper.sh>
When checking clh's binary path if valid, return error even
though the error is not a IsNotExist error.
And add errors to log filed when errors occurred.
Fixes: #2208
Signed-off-by: bin <bin@hyper.sh>
Removes custom trace functions defined across the repo and creates
a single trace function in a new katatrace package. Also moves
span tag management into this package and provides a function to
dynamically add a tag at runtime, such as a container id, etc.
Fixes#1162
Signed-off-by: Benjamin Porter <bporter816@gmail.com>
We'd better stop the virtiofsd specifically after stop qemu,
instead of depending on the qemu's termination to notify virtiofsd
to exit.
Fixes: #2211
Signed-off-by: fupan.lfp <fupan.lfp@antgroup.com>
For some reason our static check started to get opinionated about code
that's been there for ages.
One of the suggestions is to improve:
```
INFO: Running golangci-lint on /home/fidencio/go/src/github.com/kata-containers/kata-containers/src/runtime/containerd-shim-v2
utils_test.go:76:36: S1039: unnecessary use of fmt.Sprintf (gosimple)
testDir, err = ioutil.TempDir("", fmt.Sprintf("shimV2-"))
```
And that's what this PR is about.
Fixes: #2204
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
For some reason our static check started to get opinionated about code
that's been there for ages.
One of the suggestions is to improve:
```
INFO: Running golangci-lint on /home/fidencio/go/src/github.com/kata-containers/kata-containers/src/runtime/virtcontainers/pkg/agent/protocols/client
client.go:431:2: S1017: should replace this `if` statement with an unconditional `strings.TrimPrefix` (gosimple)
if strings.HasPrefix(sock, "mock:") {
```
And that's what this PR is about.
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
First of all, cpuset.go just comes from kubernetes and we shouldn't be
doing much with this file apart from updating it every now and then
(but that's material for another PR).
Right now, due to some change on the static checks we use as part of our
CI, we started getting issues as:
```
INFO: Running golangci-lint on /home/fidencio/go/src/github.com/kata-containers/kata-containers/src/runtime/virtcontainers/pkg/cpuset
cpuset.go:60:2: SA4005: ineffective assignment to field Builder.done (staticcheck)
b.done = true
```
For those, let's just ignore the lint and move on.
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
Since we only send an shutdown qmp command to qemu when do
stopSandbox, and didn't wait until qemu process's exit, thus
we'd better to make sure it had exited when shimv2 terminated.
Thus here to do the last cleanup of the hypervisor.
Fixes: #2198
Signed-off-by: fupan.lfp <fupan.lfp@antgroup.com>
It looks like the version check for cloud hypervisor (clh) was added
initially when clh was actively evolving its API. We no longer need the
version check as clh API has been fairly stable for its recent releases.
Fixes: #1991
Signed-off-by: Bo Chen <chen.bo@intel.com>
Pass span context through ttRPC metadata, that
agent can get parent from the context to create
new sub-spans.
Fixes: #1968
Signed-off-by: bin <bin@hyper.sh>
Use common functions wrapping logic of getting values
from annotations, parsing bool/uint32/uint64 and setting
to struct fields.
Fixes: #2082
Signed-off-by: bin <bin@hyper.sh>
Since the rpc spec used an interface to represen the ErrnoRet,
thus the transform function of OCItoGRPC should take care of
this case.
Depends-on: github.com/kata-containers/tests#3629
Fixes: #1441
Signed-off-by: fupan.lfp <fupan.lfp@antgroup.com>
Fix the mismatch bring by the upgrading of vendor of containerd,
cgroup and runtime spec.
Fixes: #1441
Signed-off-by: fupan.lfp <fupan.lfp@antgroup.com>
Since the latest containerd's shimv2 had changed the socket
from abstract unix socket to path unix socket, thus we'd
better to update the vendor to match with the latest containerd.
containerd from v1.3.9, v1.4.3 and v1.5.0 used the path unix socket
instead of abstract socket, thus kata wouldn's support the
containerd's version older than them.
Fixes: #1441
short logs:
15d9703d6 Remove ARM64 releases from release notes
5d2e8e86d Revert "Release artifacts for Linux ARM64"
7942ae68b Revert "Specify seccomp target arch for CC"
3187b6dc8 tests: Adds consumed memory stats test
969ec8949 Specify seccomp target arch for CC
c19b7b64d RELEASES.md: recommend alternatives for deprecated
features
8a62aa1c3 Deprecate built-in aufs snapshotter
4e7915f80 CI: allow Go 1.13 for Docker/Moby compatibility
8e589e873 Vagrantfile: update to Fedora 34
5847340a7 tests: Refactors container image usage
9f43eade6 Prepare v1.5.0-rc.3 release notes
4c7b960cb prow needs some additional setup for docker buildx
2e4c1d4b7 Use the multi-arch version of the test images
4e00c4b65 integration tests needs lsof
177273680 Add script to build test images
1b5d59dfe Add multi-arch support for test images
78e529727 add integration tests
2b0e6cdd4 Separate jobs for build and test for openlab/arm64
cdd075853 Release artifacts for Linux ARM64
efcb18742 Add unit tests for PID NamespaceMode_TARGET validation
b48f27df6 Support PID NamespaceMode_TARGET
909660ea9 process: use the unbuffered channel as the done signal
0f332dadd Update cgroups for regenerated protos
391b123a5 adds quiet option for ref
ab1654d0e Fix PushHandler cannot push image that contains duplicated
blobs
00f8d32ef add not found debug out for check cmd; update usage
55734b1c5 Prepare 1.5.0-rc.2 release notes
3ef337ae3 Update containerd vendors to tags
fbe1e140f Update Go to 1.16.3
c1d1edbad gha: use sudo -E in some places to prevent dropping
env-vars
7966a6652 Cleanup code
5d79d3adb go.mod: update kubernetes to v1.20.6
1c03c377e go.mod: github.com/containerd/fifo v1.0.0
12a2a2108 go.mod: github.com/google/uuid v1.2.0
3292ea586 pkg/seccomp: use sync.Once to speed up IsEnabled
00b5c99b1 pkg/seccomp: simplify IsEnabled, update doc
6dd29c25f go.mod: github.com/containerd/aufs
330a2a809 go.mod: github.com/containerd/zfs
34780d67a runtime/shim: check the namespace flag first
c3dde8c4b freebsd: add zfs to the default plugins
b431fe4fc freebsd: don't run shim delete in deleted dir
1f4192daf freebsd: exclude v1 runtimes
cb1580937 metadata: improve deleting a non-empty namespace's error
message
5bf84034d Remove junit test result processor
b83d04f91 Add variable names to runtime's interface definitions
993b86399 Add shim start opts
9e576b889 Optimize backoff
5c02688b5 converter: use OpenWriter helper function
fcf3b275f Add lock for ListPids
fdb76f55d Fix backword-compatibility issue of non-versioned config
file
d21fe4625 adds log for each failed host and status not found on host
8a4cbabc6 Reimport windows layers when comitting snapshots
2de38a926 fix(windows): create debug npipe failure
41fc516a2 docs/rootless.md: recommend "easy way" over "hard way"
864a3322b go.mod: github.com/containerd/go-cni v1.0.2
ee34caccb go.mod: github.com/Microsoft/go-winio v0.4.17
d478676d3 go.mod: github.com/containerd/imgcrypt v1.1.1
1dd45d51c go.mod: github.com/containerd/typeurl v1.0.2
abd4be07a fix the 404 url
978ebbef6 Prepare 1.5.0-rc.1 release
ce116d4c5 go.mod: github.com/containerd/imgcrypt
v1.1.1-0.20210412181126-0bed51b9522c
0550c3233 containerd-stress: add snapshotter option for stress test
to use
8a04bd052 address recent runtimes config confusion
c4778fe1b go.mod: github.com/containernetworking/plugins v0.9.1
5ce35ac39 devmapper: log pool status when mkfs fails
75097b8ca hcsshim seems to have been updated
9ad087947 Switch all our tests to version 2
e96d2a5d9 Revert "remove two very old no longer used runtime
options"
14f357b90 CI: update crun to 0.19
294331060 go.mod: github.com/containerd/console v1.0.2
bb6c0c2de Add more bolt utils
0ad8c0a16 Decouple shim start from task creation
c7504987e Implement windowsDiff.Compare via hcsshim/pkg/ociwclayer
a64a76846 Replace inline applyWindowsLayer using hcsshim
149fa366f Don't tease the logger with a %-less format string
b399e2ef6 Don't lose Compare failure if aborting diff upload fails
36bf3f0e8 go.mod: github.com/Microsoft/hcsshim v0.8.16
8e1a8ecd8 Prepare v1.5.0-rc.0
45df696bf Fix return event publishing error
4bc8f692f optimize cri redirect logs
9bc8d63c9 cri/server: use containerd/oci instead of
libcontainer/devices
dd16b006e merge in the move to the new options type
9144ce967 shows our runc.v2 default options in the containerd
default config
3d20fa930 fix TestSetOOMScoreBoundaries
4d4117415 Change CRI config runtime options type
21ebeef74 integration: use busybox:1.32.0 since latest is
unavailable
f9bcf4a8a add section link
d4be6aa8f rm mirror defaults; doc registry deprecations
7bb73da6b runtime/v2/shim: remove unused SetScore() and remove
sys.OOMScoreMaxKillable
91e7d21ee sys: add AdjustOOMScore() utility
44240116a sys: add boundary checks to SetOOMScore()
ace1912bb sys: use assert for error checks in OOM tests
6e7271522 sys: add missing pre-condition checks in tests
badd60d3f sys: un-export runningPrivileged(), remove
runningUnprivileged()
21a175860 go.mod github.com/klauspost/compress v1.11.13
58c5fd09e re-enable cri test
da998c81e move to gcr.io/k8s-staging-cri-tools test images
8ba8533bd pkg/cri/opts.WithoutRunMount -> oci.WithoutRunMount
92ea98eda cri-cni-release: add imgcrypt binaries (v1.1.0)
4c1fa5719 remotes/docker: Only return "already exists" on push when
the upload was successful
0186a329e remove two very old no longer used runtime options
58a07754a Temporarily disable cri-tools critest
7ae0a60fb Add OCI ref.name to unique key in remotes handler
5ada2f74a Keep host order as defined in TOML file
d9ff8ebef support multi-arch images for windows via ctr
af1e2af72 ci: upload junit formatted test results
6866b36ab Add workaround to keep docker hosts structs private
c54d92c79 image: use generic decompressor for calculating DiffID
1faca349e integration/client: rename package to "client"
6fc9e4500 synchronize replace rules in integration/client go.mod
with main go.mod
9e19a2984 Fix hosts test on Windows
3f406d4af Cleanup vendor
d56b49c13 Rewrite Docker hosts parser
e1f51ba73 Use os.File#Seek() to get the size of a block device
ddd4298a1 Migrate current TOML code to github.com/pelletier/go-toml
499c2f7d4 Vendor github.com/pelletier/go-toml
61c749036 integration/util: remove dependency on k8s.io/klog/v2
d9765f7bf Extend default timeout for nested VM integration run
5e94745f2 ctr: add --user for task exec
f8c2f0475 remotes/ctr: allow to limit max concurrent uploads like
downloads
4674ad7be Ignore some tests on darwin
55450e773 Run unit tests on CI for MacOS
311e326a1 Add CI job to cross compile all the things
10a498c7c Update go-winio to fix compile error on armv7
1a9c6f557 Revendor zfs to to fix integer overflow
1fd3d12f9 `go mod tidy` the client integration test module
da7d96ba3 Clean up WCOW layers after tests in the correct order
9ad87b9ba adds critools-version
72b7f4bab task: allow checkpoint on pause state
e4b9b1038 Make CRI registry docs more clear
ec4d7736d Increase timeout for linux integration tests
eb7c7c71e Fix oom tests on non Linux
708299ca4 Move RunningInUserNS() to its own package
0886ceaea Fix reference ordering in CRI image store
bf9db47e8 add caller info to the testHook
305b42583 use happy-eyeballs for port-forwarding
22ef69d77 Support HTTP debug in ctr
01765d097 night ci fix: add packages for ubuntu 20.04
8cdc1f13b go.mod: github.com/containerd/zfs
v0.0.0-20210322090317-0e92c2247fb7
30e1e66e5 runtime/v2: Fix defer cleanup
33776ada0 Use specific image for user namespaces tests
7704fe72d Specifically mention "mkfs.ext4" on the error from the
command
1410220d8 Fix error log when copy file
fe787efa2 Fix error log when kill shim
8d8c15ca5 contentproxy: ensure grpc stream is closed on commit
6e343f25e Switch test image to a non rate-limited manifest list
9fdc96c09 runtime/v2: add comment for checkCopyShimLogError
24602e7a9 change default runtime for containerd-stress app
8731888ec Re-enable CRIU tests by not using overlayfs snapshotter
b520428b5 Fix CRIU
4e76bcf06 gofmt -s -w all the things
569023fd5 go.mod: github.com/containerd/nri
v0.0.0-20210316161719-dbaa18c31c14
0e1f59e89 go.mod: github.com/containerd/zfs
v0.0.0-20210315114300-dde8f0fda960
ffff68866 upgrade pause image to 3.5 for non-root
88d3881e1 go.mod: github.com/containerd/fifo
v0.0.0-20210316144830-115abcc95a1d
a22c43fa4 go.mod: github.com/containerd/aufs
v0.0.0-20210316121734-20793ff83c97
f6f861736 go.mod: github.com/containerd/btrfs
v0.0.0-20210316141732-918d888fb676
460b35236 go.mod: kubernetes v1.20.4
5e484c961 runtime/v2/runc: fix the defer cleanup of the NewContainer
e6086d9c0 Prepare release notes for v1.5.0-beta.4
34b7a5f09 Update mailmap
ba8f9845e move overlay-checks to an overlayutils package
7776e5ef2 Support adding devices by dir
d895118c7 runtime/v2/runc: fix leaking socket path
a76cefd12 plugin status should be skip, not error
766e7953a Change dgst to digest in debug
4e8b2f309 rootfs: fix the error handling of the createInitLayer
d3ad7f390 cmd/ctr: use e.g. in the command usage
231bbdc37 cmd/ctr: fix export command
ecb881e5e add imgcrypt stream processors to the default config
ac2726e12 cmd/containerd: deduplicate config*.go
9a7ca39cb defaults: add DefaultConfigDir
8f863afd3 Use net.IP.IsLoopback() to match loopback addresses
eabd9b98b runtime: ignore file-already-closed error if dead shim
Signed-off-by: fupan.lfp <fupan.lfp@antgroup.com>
To workaround virtiofs' lack of inotify support, we'll special case
particular mounts which are typically watched, and pass on information
to the agent so it can ensure that the mount presented to the container
is indeed watchable (see applicable agent commit).
This commit will:
- identify watchable mounts based on file count and mount source
- create a watchable-bind storage object for these mounts to
communicate intent to the agent
- update the OCI spec to take the updated watchable mount source into account
Unit tests added and updated for the newly introduced
functionality/functions.
Signed-off-by: Eric Ernst <eric_ernst@apple.com>
There's no reason to pass the paths; they can be
determined when they are actually used.
Let's make the return values more comparable to the other mount handling
functions (we'll add storage object in future commit), and pass the mount maps as
function parameters.
...No functional changes here...
Signed-off-by: Eric Ernst <eric_ernst@apple.com>
Architectures that do not support memory hotplugging will fail when
memory limits are set because that amount is hotplugged. Issue a warning
instead. The long-term solution is virtio-mem.
Fixes: #1412
Signed-off-by: Jakob Naucke <jakob.naucke@ibm.com>
After create an container/exec successfully, containerd
would wait it immediately, and if start it failed, there
is no chance to send value to exitCh, thus the wait goroutine
would blocked for ever and had no chance to exit.
Fixes: #2087
Signed-off-by: fupan.lfp <fupan.lfp@antgroup.com>
