Files
Enterprise-Knowledge-Base/internal/im/feishu/factory.go
T
养乐多 baada32cb6 fix(im/feishu): actually close WebSocket long connection when channel is stopped
Deleting or disabling a Feishu channel left its WebSocket long connection
alive: the Lark SDK's Start blocks on a bare select{} and never observes
ctx, and v3.6.1 exposed no public Close(), so the wsCancel() cleanup was a
no-op. The socket stayed open, the SDK kept auto-reconnecting, and Feishu
kept pushing messages — causing cross-environment message duplication when
the same bot was bound elsewhere.

Fix in three layers:

1. Connection teardown
   - Upgrade oapi-sdk-go v3.6.1 -> v3.9.7 (adds Client.Close()).
   - Add LongConnClient.Close() wrapping wsClient.Close(); factory's cleanup
     now calls Close() first, then wsCancel() as a belt-and-braces fallback.

2. Multi-instance coordination
   - wsLeaderRetryLoop: after acquiring leadership, re-check the DB before
     starting; abort if the channel was deleted/disabled while waiting.
   - wsLeaderRenewLoop: on each renewal, re-check the DB; step down if the
     channel was deleted/disabled on another instance (HTTP request may have
     been served elsewhere), so the leader doesn't keep the connection open
     until process restart.

3. Robustness
   - DB checks distinguish gorm.ErrRecordNotFound (channel truly gone ->
     stop/abort) from transient DB errors (skip this round, retry next tick)
     so a DB hiccup can't cause a spurious stop or permanent leader-abort.
   - startChannelInternal now stops any existing channelState for the same
     ID before overwriting, making it idempotent against concurrent start
     paths and preventing adapter/long-connection leaks.

No interface changes; webhook and long-poll modes unaffected.

Co-Authored-By: Claude <noreply@anthropic.com>
2026-07-06 23:10:45 +08:00

61 lines
2.0 KiB
Go

package feishu
import (
"context"
"fmt"
"github.com/Tencent/WeKnora/internal/im"
"github.com/Tencent/WeKnora/internal/logger"
)
// NewFactory returns an im.AdapterFactory for Feishu (Lark) channels.
// The HTTP adapter is always created (needed for SendReply in both modes);
// "websocket" mode additionally runs a long-connection event stream.
func NewFactory() im.AdapterFactory {
return func(factoryCtx context.Context, channel *im.IMChannel, msgHandler func(context.Context, *im.IncomingMessage) error) (im.Adapter, context.CancelFunc, error) {
creds, err := im.ParseCredentials(channel.Credentials)
if err != nil {
return nil, nil, fmt.Errorf("parse feishu credentials: %w", err)
}
appID := im.GetString(creds, "app_id")
appSecret := im.GetString(creds, "app_secret")
verificationToken := im.GetString(creds, "verification_token")
encryptKey := im.GetString(creds, "encrypt_key")
// Always create the HTTP adapter (needed for SendReply in both modes)
adapter := NewAdapter(appID, appSecret, verificationToken, encryptKey)
mode := im.ResolveMode(channel, "websocket")
switch mode {
case "webhook":
return adapter, nil, nil
case "websocket":
client := NewLongConnClient(appID, appSecret, msgHandler)
wsCtx, wsCancel := context.WithCancel(context.Background())
go func() {
if err := client.Start(wsCtx); err != nil && wsCtx.Err() == nil {
logger.Errorf(context.Background(), "[IM] Feishu long connection stopped for channel %s: %v", channel.ID, err)
}
}()
// stop tears down the connection for real. wsCancel alone is a
// no-op: the Feishu SDK's Start blocks on select{} and never
// observes ctx, so we must call Close() to actually close the
// socket and disable auto-reconnect. wsCancel is kept as a
// belt-and-braces fallback for the start goroutine.
stop := func() {
client.Close()
wsCancel()
}
return adapter, stop, nil
default:
return nil, nil, fmt.Errorf("unknown Feishu mode: %s", mode)
}
}
}