github/kubernetes
4
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Better back off delays and connection timeout to talk to containerd

Browse Source 
Set up params similar to what we do in cadvisor:
e9068e3273/container/containerd/client.go (L59-L61)

Signed-off-by: Davanum Srinivas <davanum@gmail.com>
This commit is contained in:
Davanum Srinivas 2023-06-25 16:23:44 -04:00
parent ba6d2674ca
commit f7239e4095
No known key found for this signature in database
GPG Key ID: 80D83A796103BF59
2 changed files with 28 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
pkg/kubelet/cri/remote/remote_image.go
View File

@ -56,11 +56,7 @@ func NewRemoteImageService(endpoint string, connectionTimeout time.Duration, tp
ctx, cancel := context.WithTimeout(context.Background(), connectionTimeout)
defer cancel()
dialOpts := []grpc.DialOption{
grpc.WithConnectParams(grpc.ConnectParams{
Backoff: backoff.DefaultConfig,
}),
}
var dialOpts []grpc.DialOption
dialOpts = append(dialOpts,
grpc.WithTransportCredentials(insecure.NewCredentials()),
grpc.WithContextDialer(dialer),
@ -77,6 +73,16 @@ func NewRemoteImageService(endpoint string, connectionTimeout time.Duration, tp
grpc.WithStreamInterceptor(otelgrpc.StreamClientInterceptor(tracingOpts...)))
}
connParams := grpc.ConnectParams{
Backoff: backoff.DefaultConfig,
}
connParams.MinConnectTimeout = minConnectionTimeout
connParams.Backoff.BaseDelay = baseBackoffDelay
connParams.Backoff.MaxDelay = maxBackoffDelay
dialOpts = append(dialOpts,
grpc.WithConnectParams(connParams),
)
conn, err := grpc.DialContext(ctx, addr, dialOpts...)
if err != nil {
klog.ErrorS(err, "Connect remote image service failed", "address", addr)

22
pkg/kubelet/cri/remote/remote_runtime.go
View File

@ -56,6 +56,11 @@ type remoteRuntimeService struct {
const (
// How frequently to report identical errors
identicalErrorDelay = 1 * time.Minute
// connection parameters
maxBackoffDelay = 3 * time.Second
baseBackoffDelay = 100 * time.Millisecond
minConnectionTimeout = 5 * time.Second
)
// CRIVersion is the type for valid Container Runtime Interface (CRI) API
@ -80,11 +85,7 @@ func NewRemoteRuntimeService(endpoint string, connectionTimeout time.Duration, t
ctx, cancel := context.WithTimeout(context.Background(), connectionTimeout)
defer cancel()
dialOpts := []grpc.DialOption{
grpc.WithConnectParams(grpc.ConnectParams{
Backoff: backoff.DefaultConfig,
}),
}
var dialOpts []grpc.DialOption
dialOpts = append(dialOpts,
grpc.WithTransportCredentials(insecure.NewCredentials()),
grpc.WithContextDialer(dialer),
@ -100,6 +101,17 @@ func NewRemoteRuntimeService(endpoint string, connectionTimeout time.Duration, t
grpc.WithUnaryInterceptor(otelgrpc.UnaryClientInterceptor(tracingOpts...)),
grpc.WithStreamInterceptor(otelgrpc.StreamClientInterceptor(tracingOpts...)))
}
connParams := grpc.ConnectParams{
Backoff: backoff.DefaultConfig,
}
connParams.MinConnectTimeout = minConnectionTimeout
connParams.Backoff.BaseDelay = baseBackoffDelay
connParams.Backoff.MaxDelay = maxBackoffDelay
dialOpts = append(dialOpts,
grpc.WithConnectParams(connParams),
)
conn, err := grpc.DialContext(ctx, addr, dialOpts...)
if err != nil {
klog.ErrorS(err, "Connect remote runtime failed", "address", addr)