fix(sync): use per-cluster TLS for HTTP volume connections in filer.sync (#8974)

* fix(sync): use per-cluster TLS for HTTP volume connections in filer.sync (#8965)

When filer.sync runs with -a.security and -b.security flags, only gRPC
connections received per-cluster TLS configuration. HTTP clients for
volume server reads and uploads used a global singleton with the default
security.toml, causing TLS verification failures when clusters use
different self-signed certificates.

Load per-cluster HTTPS client config from the security files and pass
dedicated HTTP clients to FilerSource (for downloads) and FilerSink
(for uploads) so each direction uses the correct cluster's certificates.

* fix(sync): address review feedback for per-cluster HTTP TLS

- Add insecure_skip_verify support to NewHttpClientWithTLS and read it
  from per-cluster security config via https.client.insecure_skip_verify
- Error on partial mTLS config (cert without key or vice versa)
- Add nil-check for client parameter in DownloadFileWithClient
- Document SetUploader as init-only (same pattern as SetChunkConcurrency)
This commit is contained in:
Chris Lu authored and GitHub committed 2026-04-07 14:11:44 -07:00
1 parent d50889002b
commit 2919bb27e5
8 files changed
+179 -8

No files matched your search

+31 -3
View File
@@ -15,12 +15,14 @@ import (
"github.com/seaweedfs/seaweedfs/weed/pb/filer_pb"
"github.com/seaweedfs/seaweedfs/weed/replication"
"github.com/seaweedfs/seaweedfs/weed/replication/sink"
"github.com/seaweedfs/seaweedfs/weed/operation"
"github.com/seaweedfs/seaweedfs/weed/replication/sink/filersink"
"github.com/seaweedfs/seaweedfs/weed/replication/source"
"github.com/seaweedfs/seaweedfs/weed/security"
statsCollect "github.com/seaweedfs/seaweedfs/weed/stats"
"github.com/seaweedfs/seaweedfs/weed/util"
"github.com/seaweedfs/seaweedfs/weed/util/grace"
util_http_client "github.com/seaweedfs/seaweedfs/weed/util/http/client"
"github.com/seaweedfs/seaweedfs/weed/util/wildcard"
"google.golang.org/grpc"
)
@@ -164,6 +166,21 @@ func runFilerSynchronize(cmd *Command, args []string) bool {
}
}
// per-cluster HTTPS clients for volume server connections
var httpClientA, httpClientB *util_http_client.HTTPClient
if *syncOptions.aSecurity != "" {
var err error
if httpClientA, err = security.LoadHTTPClientFromFile(*syncOptions.aSecurity); err != nil {
glog.Fatalf("load HTTPS client config for filer A: %v", err)
}
}
if *syncOptions.bSecurity != "" {
var err error
if httpClientB, err = security.LoadHTTPClientFromFile(*syncOptions.bSecurity); err != nil {
glog.Fatalf("load HTTPS client config for filer B: %v", err)
}
}
grace.SetupProfiling(*syncCpuProfile, *syncMemProfile)
filerA := pb.ServerAddress(*syncOptions.filerA)
@@ -238,7 +255,9 @@ func runFilerSynchronize(cmd *Command, args []string) bool {
*syncOptions.bDoDeleteFiles,
aFilerSignature,
bFilerSignature,
&syncStateA2B)
&syncStateA2B,
httpClientA,
httpClientB)
if err != nil {
glog.Errorf("sync from %s to %s: %v", *syncOptions.filerA, *syncOptions.filerB, err)
time.Sleep(1747 * time.Millisecond)
@@ -279,7 +298,9 @@ func runFilerSynchronize(cmd *Command, args []string) bool {
*syncOptions.aDoDeleteFiles,
bFilerSignature,
aFilerSignature,
&syncStateB2A)
&syncStateB2A,
httpClientB,
httpClientA)
if err != nil {
glog.Errorf("sync from %s to %s: %v", *syncOptions.filerB, *syncOptions.filerA, err)
time.Sleep(2147 * time.Millisecond)
@@ -308,7 +329,8 @@ func initOffsetFromTsMs(grpcDialOption grpc.DialOption, targetFiler pb.ServerAdd
}
func doSubscribeFilerMetaChanges(clientId int32, clientEpoch int32, sourceGrpcDialOption grpc.DialOption, sourceFiler pb.ServerAddress, sourcePath string, sourceExcludePaths []string, sourceReadChunkFromFiler bool, targetGrpcDialOption grpc.DialOption, targetFiler pb.ServerAddress, targetPath string,
replicationStr, collection string, ttlSec int, sinkWriteChunkByFiler bool, diskType string, debug bool, concurrency int, chunkConcurrency int, doDeleteFiles bool, sourceFilerSignature int32, targetFilerSignature int32, statePtr *atomic.Pointer[syncState]) error {
replicationStr, collection string, ttlSec int, sinkWriteChunkByFiler bool, diskType string, debug bool, concurrency int, chunkConcurrency int, doDeleteFiles bool, sourceFilerSignature int32, targetFilerSignature int32, statePtr *atomic.Pointer[syncState],
sourceHttpClient *util_http_client.HTTPClient, sinkHttpClient *util_http_client.HTTPClient) error {
// if first time, start from now
// if has previously synced, resume from that point of time
@@ -323,9 +345,15 @@ func doSubscribeFilerMetaChanges(clientId int32, clientEpoch int32, sourceGrpcDi
filerSource := &source.FilerSource{}
filerSource.DoInitialize(sourceFiler.ToHttpAddress(), sourceFiler.ToGrpcAddress(), sourcePath, sourceReadChunkFromFiler)
filerSource.SetGrpcDialOption(sourceGrpcDialOption)
if sourceHttpClient != nil {
filerSource.SetHttpClient(sourceHttpClient)
}
filerSink := &filersink.FilerSink{}
filerSink.DoInitialize(targetFiler.ToHttpAddress(), targetFiler.ToGrpcAddress(), targetPath, replicationStr, collection, ttlSec, diskType, targetGrpcDialOption, sinkWriteChunkByFiler)
filerSink.SetChunkConcurrency(chunkConcurrency)
if sinkHttpClient != nil {
filerSink.SetUploader(operation.NewUploaderWithHttpClient(sinkHttpClient))
}
filerSink.SetSourceFiler(filerSource)
persistEventFn := genProcessFunction(sourcePath, targetPath, sourceExcludePaths, nil, nil, nil, filerSink, doDeleteFiles, debug)