-
Notifications
You must be signed in to change notification settings - Fork 8
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
* use dynstream in maintainer
- Loading branch information
Showing
14 changed files
with
1,462 additions
and
363 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,187 @@ | ||
// Copyright 2024 PingCAP, Inc. | ||
// | ||
// Licensed under the Apache License, Version 2.0 (the "License"); | ||
// you may not use this file except in compliance with the License. | ||
// You may obtain a copy of the License at | ||
// | ||
// http://www.apache.org/licenses/LICENSE-2.0 | ||
// | ||
// Unless required by applicable law or agreed to in writing, software | ||
// distributed under the License is distributed on an "AS IS" BASIS, | ||
// See the License for the specific language governing permissions and | ||
// limitations under the License. | ||
|
||
package maintainer | ||
|
||
import ( | ||
"github.com/flowbehappy/tigate/heartbeatpb" | ||
"github.com/flowbehappy/tigate/pkg/common" | ||
"github.com/flowbehappy/tigate/pkg/messaging" | ||
"github.com/flowbehappy/tigate/scheduler" | ||
"github.com/pingcap/log" | ||
"go.uber.org/zap" | ||
"time" | ||
) | ||
|
||
// Bootstrapper handle the logic of the maintainer startup | ||
// when a maintainer is started, it must wait of node to reported their managed dispatchers | ||
// only all dispatcher has reported its status, maintainer can schedule tables | ||
// maintainer collects all working dispatchers and | ||
type Bootstrapper struct { | ||
// changefeedID is the log identifier | ||
changefeedID string | ||
// bootstrap identify is the bootstrapper is already bootstrapped | ||
bootstrapped bool | ||
|
||
nodes map[common.NodeID]*NodeStatus | ||
newBootstrapMsg scheduler.NewBootstrapFn | ||
} | ||
|
||
// NewBootstrapper create a new bootstrap for a changefeed maintainer | ||
func NewBootstrapper(cfID string, newBootstrapMsg scheduler.NewBootstrapFn) *Bootstrapper { | ||
return &Bootstrapper{ | ||
changefeedID: cfID, | ||
nodes: make(map[common.NodeID]*NodeStatus), | ||
bootstrapped: false, | ||
newBootstrapMsg: newBootstrapMsg, | ||
} | ||
} | ||
|
||
// HandleBootstrapResponse cache the message reported remote node | ||
func (b *Bootstrapper) HandleBootstrapResponse( | ||
from messaging.ServerId, | ||
msg *heartbeatpb.MaintainerBootstrapResponse) map[common.NodeID]*heartbeatpb.MaintainerBootstrapResponse { | ||
nodeID := common.NodeID(from) | ||
node, ok := b.nodes[nodeID] | ||
if !ok { | ||
log.Warn("node is not found, ignore", | ||
zap.String("changefeed", b.changefeedID), | ||
zap.String("from", nodeID)) | ||
return nil | ||
} | ||
node.cachedBootstrapResp = msg | ||
node.state = NodeStateInitialized | ||
return b.fistBootstrap() | ||
} | ||
|
||
// HandleNewNodes add node to bootstrapper and return rpc messages that need to be sent to remote node | ||
func (b *Bootstrapper) HandleNewNodes(nodes map[common.NodeID]*common.NodeInfo) []*messaging.TargetMessage { | ||
msgs := make([]*messaging.TargetMessage, 0, len(nodes)) | ||
for id, info := range nodes { | ||
if _, ok := b.nodes[info.ID]; !ok { | ||
// A new server. | ||
b.nodes[info.ID] = NewNodeStatus(info) | ||
log.Info("find a new server", | ||
zap.String("changefeed", b.changefeedID), | ||
zap.String("captureAddr", info.AdvertiseAddr), | ||
zap.String("server", info.ID)) | ||
msgs = append(msgs, b.newBootstrapMsg(id)) | ||
b.nodes[info.ID].lastBootstrapTime = time.Now() | ||
} | ||
} | ||
return msgs | ||
} | ||
|
||
// HandleRemoveNodes remove node from bootstrapper, | ||
// finished bootstrap if all node are initialized after these node removed | ||
// return cached bootstrap | ||
func (b *Bootstrapper) HandleRemoveNodes(nodeIds []string) map[common.NodeID]*heartbeatpb.MaintainerBootstrapResponse { | ||
for _, id := range nodeIds { | ||
status, ok := b.nodes[id] | ||
if ok { | ||
delete(b.nodes, id) | ||
log.Info("remove node from bootstrap", | ||
zap.String("changefeed", b.changefeedID), | ||
zap.Int("status", int(status.state)), | ||
zap.String("id", id)) | ||
} else { | ||
log.Info("node is node tracked by bootstrap", | ||
zap.String("changefeed", b.changefeedID), | ||
zap.String("id", id)) | ||
} | ||
} | ||
return b.fistBootstrap() | ||
} | ||
|
||
// ResendBootstrapMessage return rpc message that need to be resent | ||
func (b *Bootstrapper) ResendBootstrapMessage() []*messaging.TargetMessage { | ||
var msgs []*messaging.TargetMessage | ||
if !b.CheckAllNodeInitialized() { | ||
for id, node := range b.nodes { | ||
if node.state == NodeStateUninitialized && | ||
time.Since(node.lastBootstrapTime) > time.Millisecond*500 { | ||
msgs = append(msgs, b.newBootstrapMsg(id)) | ||
node.lastBootstrapTime = time.Now() | ||
} | ||
} | ||
} | ||
return msgs | ||
} | ||
|
||
// GetAllNodes return all nodes the tracked by bootstrapper, the returned value must not be modified | ||
func (b *Bootstrapper) GetAllNodes() map[common.NodeID]*NodeStatus { | ||
return b.nodes | ||
} | ||
|
||
// CheckAllNodeInitialized check if all server is initialized. | ||
// returns true when all server reports the bootstrap response and bootstrapped | ||
func (b *Bootstrapper) CheckAllNodeInitialized() bool { | ||
return b.bootstrapped && b.checkAllCaptureInitialized() | ||
} | ||
|
||
// return true if all node reports the bootstrap response | ||
func (b *Bootstrapper) checkAllCaptureInitialized() bool { | ||
for _, captureStatus := range b.nodes { | ||
// CaptureStateStopping is also considered initialized, because when | ||
// a server shutdown, it becomes stopping, we need to move its tables | ||
// to other captures. | ||
if captureStatus.state == NodeStateUninitialized { | ||
return false | ||
} | ||
} | ||
return len(b.nodes) != 0 | ||
} | ||
|
||
// fistBootstrap check if bootstrapper is initialized first time, | ||
// return nil is not bootstrapped or already bootstrapped before | ||
// return cached heartbeatpb.MaintainerBootstrapResponse map if it's not bootstrapped before | ||
// bootstrapper only return once | ||
func (b *Bootstrapper) fistBootstrap() map[common.NodeID]*heartbeatpb.MaintainerBootstrapResponse { | ||
// first bootstrapped time, return the cached resp and clear it | ||
if !b.bootstrapped && b.checkAllCaptureInitialized() { | ||
b.bootstrapped = true | ||
allCachedResp := make(map[common.NodeID]*heartbeatpb.MaintainerBootstrapResponse, len(b.nodes)) | ||
for _, node := range b.nodes { | ||
allCachedResp[node.node.ID] = node.cachedBootstrapResp | ||
// clear the cached data | ||
node.cachedBootstrapResp = nil | ||
} | ||
return allCachedResp | ||
} | ||
return nil | ||
} | ||
|
||
type NodeState int | ||
|
||
const ( | ||
// NodeStateUninitialized means the server status is unknown, | ||
// no bootstrap response received yet. | ||
NodeStateUninitialized NodeState = 1 | ||
// NodeStateInitialized means scheduler has received bootstrap response. | ||
NodeStateInitialized NodeState = 2 | ||
) | ||
|
||
func NewNodeStatus(node *common.NodeInfo) *NodeStatus { | ||
return &NodeStatus{ | ||
state: NodeStateUninitialized, | ||
node: node, | ||
} | ||
} | ||
|
||
// NodeStatus identify the node the need be bootstrapped | ||
type NodeStatus struct { | ||
state NodeState | ||
node *common.NodeInfo | ||
cachedBootstrapResp *heartbeatpb.MaintainerBootstrapResponse | ||
lastBootstrapTime time.Time | ||
} |
Oops, something went wrong.