This commit is contained in:
chrislu 2022-07-10 14:01:53 -07:00
parent ef93869009
commit 24291e23eb
5 changed files with 147 additions and 176 deletions

View file

@ -15,24 +15,13 @@ const (
BrokerType = "broker" BrokerType = "broker"
) )
type FilerGroup string type FilerGroupName string
type Filers struct { type DataCenter string
members map[pb.ServerAddress]*ClusterNode type Rack string
leaders *Leaders
}
type Leaders struct { type Leaders struct {
leaders [3]pb.ServerAddress leaders [3]pb.ServerAddress
} }
type DataCenter string
type Rack string
type DataCenterBrokers struct {
brokers map[Rack]*RackBrokers
}
type RackBrokers struct {
brokers map[pb.ServerAddress]*ClusterNode
}
type ClusterNode struct { type ClusterNode struct {
Address pb.ServerAddress Address pb.ServerAddress
Version string Version string
@ -41,45 +30,42 @@ type ClusterNode struct {
DataCenter DataCenter DataCenter DataCenter
Rack Rack Rack Rack
} }
type GroupMembers struct {
members map[pb.ServerAddress]*ClusterNode
leaders *Leaders
}
type ClusterNodeGroups struct {
groupMembers map[FilerGroupName]*GroupMembers
sync.RWMutex
}
type Cluster struct { type Cluster struct {
filerGroup2filers map[FilerGroup]*Filers filerGroups *ClusterNodeGroups
filersLock sync.RWMutex brokerGroups *ClusterNodeGroups
brokers map[DataCenter]*DataCenterBrokers
brokersLock sync.RWMutex
} }
func NewCluster() *Cluster { func newClusterNodeGroups() *ClusterNodeGroups {
return &Cluster{ return &ClusterNodeGroups{
filerGroup2filers: make(map[FilerGroup]*Filers), groupMembers: map[FilerGroupName]*GroupMembers{},
brokers: make(map[DataCenter]*DataCenterBrokers),
} }
} }
func (g *ClusterNodeGroups) getGroupMembers(filerGroup FilerGroupName, createIfNotFound bool) *GroupMembers {
func (cluster *Cluster) getFilers(filerGroup FilerGroup, createIfNotFound bool) *Filers { filers, found := g.groupMembers[filerGroup]
filers, found := cluster.filerGroup2filers[filerGroup]
if !found && createIfNotFound { if !found && createIfNotFound {
filers = &Filers{ filers = &GroupMembers{
members: make(map[pb.ServerAddress]*ClusterNode), members: make(map[pb.ServerAddress]*ClusterNode),
leaders: &Leaders{}, leaders: &Leaders{},
} }
cluster.filerGroup2filers[filerGroup] = filers g.groupMembers[filerGroup] = filers
} }
return filers return filers
} }
func (cluster *Cluster) AddClusterNode(ns, nodeType string, dataCenter DataCenter, rack Rack, address pb.ServerAddress, version string) []*master_pb.KeepConnectedResponse { func (m *GroupMembers) addMember(dataCenter DataCenter, rack Rack, address pb.ServerAddress, version string) *ClusterNode {
filerGroup := FilerGroup(ns) if existingNode, found := m.members[address]; found {
switch nodeType {
case FilerType:
cluster.filersLock.Lock()
defer cluster.filersLock.Unlock()
filers := cluster.getFilers(filerGroup, true)
if existingNode, found := filers.members[address]; found {
existingNode.counter++ existingNode.counter++
return nil return nil
} }
filers.members[address] = &ClusterNode{ t := &ClusterNode{
Address: address, Address: address,
Version: version, Version: version,
counter: 1, counter: 1,
@ -87,163 +73,148 @@ func (cluster *Cluster) AddClusterNode(ns, nodeType string, dataCenter DataCente
DataCenter: dataCenter, DataCenter: dataCenter,
Rack: rack, Rack: rack,
} }
return ensureFilerLeaders(filers, true, filerGroup, nodeType, address) m.members[address] = t
case BrokerType: return t
cluster.brokersLock.Lock()
defer cluster.brokersLock.Unlock()
existingDataCenterBrokers, foundDataCenter := cluster.brokers[dataCenter]
if !foundDataCenter {
existingDataCenterBrokers = &DataCenterBrokers{
brokers: make(map[Rack]*RackBrokers),
} }
cluster.brokers[dataCenter] = existingDataCenterBrokers func (m *GroupMembers) removeMember(address pb.ServerAddress) bool {
} if existingNode, found := m.members[address]; !found {
existingRackBrokers, foundRack := existingDataCenterBrokers.brokers[rack] return false
if !foundRack {
existingRackBrokers = &RackBrokers{
brokers: make(map[pb.ServerAddress]*ClusterNode),
}
existingDataCenterBrokers.brokers[rack] = existingRackBrokers
}
if existingBroker, found := existingRackBrokers.brokers[address]; found {
existingBroker.counter++
return nil
}
existingRackBrokers.brokers[address] = &ClusterNode{
Address: address,
Version: version,
counter: 1,
CreatedTs: time.Now(),
DataCenter: dataCenter,
Rack: rack,
}
return []*master_pb.KeepConnectedResponse{
{
ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{
NodeType: nodeType,
Address: string(address),
IsAdd: true,
},
},
}
case MasterType:
return []*master_pb.KeepConnectedResponse{
{
ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{
NodeType: nodeType,
Address: string(address),
IsAdd: true,
},
},
}
}
return nil
}
func (cluster *Cluster) RemoveClusterNode(ns string, nodeType string, dataCenter DataCenter, rack Rack, address pb.ServerAddress) []*master_pb.KeepConnectedResponse {
filerGroup := FilerGroup(ns)
switch nodeType {
case FilerType:
cluster.filersLock.Lock()
defer cluster.filersLock.Unlock()
filers := cluster.getFilers(filerGroup, false)
if filers == nil {
return nil
}
if existingNode, found := filers.members[address]; !found {
return nil
} else { } else {
existingNode.counter-- existingNode.counter--
if existingNode.counter <= 0 { if existingNode.counter <= 0 {
delete(filers.members, address) delete(m.members, address)
return ensureFilerLeaders(filers, false, filerGroup, nodeType, address) return true
} }
} }
case BrokerType:
cluster.brokersLock.Lock()
defer cluster.brokersLock.Unlock()
existingDataCenterBrokers, foundDataCenter := cluster.brokers[dataCenter]
if !foundDataCenter {
return nil
}
existingRackBrokers, foundRack := existingDataCenterBrokers.brokers[Rack(rack)]
if !foundRack {
return nil
}
existingBroker, found := existingRackBrokers.brokers[address]
if !found {
return nil
}
existingBroker.counter--
if existingBroker.counter <= 0 {
delete(existingRackBrokers.brokers, address)
return []*master_pb.KeepConnectedResponse{
{
ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{
NodeType: nodeType,
Address: string(address),
IsAdd: false,
},
},
}
}
return nil
case MasterType:
return []*master_pb.KeepConnectedResponse{
{
ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{
NodeType: nodeType,
Address: string(address),
IsAdd: false,
},
},
}
}
return nil
}
func (cluster *Cluster) ListClusterNode(filerGroup FilerGroup, nodeType string) (nodes []*ClusterNode) {
switch nodeType {
case FilerType:
cluster.filersLock.RLock()
defer cluster.filersLock.RUnlock()
filers := cluster.getFilers(filerGroup, false)
if filers == nil {
return
}
for _, node := range filers.members {
nodes = append(nodes, node)
}
case BrokerType:
cluster.brokersLock.RLock()
defer cluster.brokersLock.RUnlock()
for _, dcNodes := range cluster.brokers {
for _, rackNodes := range dcNodes.brokers {
for _, node := range rackNodes.brokers {
nodes = append(nodes, node)
}
}
}
case MasterType:
}
return
}
func (cluster *Cluster) IsOneLeader(filerGroup FilerGroup, address pb.ServerAddress) bool {
filers := cluster.getFilers(filerGroup, false)
if filers == nil {
return false return false
} }
return filers.leaders.isOneLeader(address)
func (g *ClusterNodeGroups) AddClusterNode(filerGroup FilerGroupName, nodeType string, dataCenter DataCenter, rack Rack, address pb.ServerAddress, version string) []*master_pb.KeepConnectedResponse {
g.Lock()
defer g.Unlock()
m := g.getGroupMembers(filerGroup, true)
if t := m.addMember(dataCenter, rack, address, version); t != nil {
return ensureGroupLeaders(m, true, filerGroup, nodeType, address)
}
return nil
}
func (g *ClusterNodeGroups) RemoveClusterNode(filerGroup FilerGroupName, nodeType string, address pb.ServerAddress) []*master_pb.KeepConnectedResponse {
g.Lock()
defer g.Unlock()
m := g.getGroupMembers(filerGroup, false)
if m == nil {
return nil
}
if m.removeMember(address) {
return ensureGroupLeaders(m, false, filerGroup, nodeType, address)
}
return nil
}
func (g *ClusterNodeGroups) ListClusterNode(filerGroup FilerGroupName) (nodes []*ClusterNode) {
g.Lock()
defer g.Unlock()
m := g.getGroupMembers(filerGroup, false)
if m == nil {
return nil
}
for _, node := range m.members {
nodes = append(nodes, node)
}
return
}
func (g *ClusterNodeGroups) IsOneLeader(filerGroup FilerGroupName, address pb.ServerAddress) bool {
g.Lock()
defer g.Unlock()
m := g.getGroupMembers(filerGroup, false)
if m == nil {
return false
}
return m.leaders.isOneLeader(address)
}
func NewCluster() *Cluster {
return &Cluster{
filerGroups: newClusterNodeGroups(),
brokerGroups: newClusterNodeGroups(),
}
} }
func ensureFilerLeaders(filers *Filers, isAdd bool, filerGroup FilerGroup, nodeType string, address pb.ServerAddress) (result []*master_pb.KeepConnectedResponse) { func (cluster *Cluster) getGroupMembers(filerGroup FilerGroupName, nodeType string, createIfNotFound bool) *GroupMembers {
switch nodeType {
case FilerType:
return cluster.filerGroups.getGroupMembers(filerGroup, createIfNotFound)
case BrokerType:
return cluster.brokerGroups.getGroupMembers(filerGroup, createIfNotFound)
}
return nil
}
func (cluster *Cluster) AddClusterNode(ns, nodeType string, dataCenter DataCenter, rack Rack, address pb.ServerAddress, version string) []*master_pb.KeepConnectedResponse {
filerGroup := FilerGroupName(ns)
switch nodeType {
case FilerType:
return cluster.filerGroups.AddClusterNode(filerGroup, nodeType, dataCenter, rack, address, version)
case BrokerType:
return cluster.brokerGroups.AddClusterNode(filerGroup, nodeType, dataCenter, rack, address, version)
case MasterType:
return []*master_pb.KeepConnectedResponse{
{
ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{
NodeType: nodeType,
Address: string(address),
IsAdd: true,
},
},
}
}
return nil
}
func (cluster *Cluster) RemoveClusterNode(ns string, nodeType string, address pb.ServerAddress) []*master_pb.KeepConnectedResponse {
filerGroup := FilerGroupName(ns)
switch nodeType {
case FilerType:
return cluster.filerGroups.RemoveClusterNode(filerGroup, nodeType, address)
case BrokerType:
return cluster.brokerGroups.RemoveClusterNode(filerGroup, nodeType, address)
case MasterType:
return []*master_pb.KeepConnectedResponse{
{
ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{
NodeType: nodeType,
Address: string(address),
IsAdd: false,
},
},
}
}
return nil
}
func (cluster *Cluster) ListClusterNode(filerGroup FilerGroupName, nodeType string) (nodes []*ClusterNode) {
switch nodeType {
case FilerType:
return cluster.filerGroups.ListClusterNode(filerGroup)
case BrokerType:
return cluster.brokerGroups.ListClusterNode(filerGroup)
case MasterType:
}
return
}
func (cluster *Cluster) IsOneLeader(filerGroup FilerGroupName, nodeType string, address pb.ServerAddress) bool {
switch nodeType {
case FilerType:
return cluster.filerGroups.IsOneLeader(filerGroup, address)
case BrokerType:
return cluster.brokerGroups.IsOneLeader(filerGroup, address)
case MasterType:
}
return false
}
func ensureGroupLeaders(m *GroupMembers, isAdd bool, filerGroup FilerGroupName, nodeType string, address pb.ServerAddress) (result []*master_pb.KeepConnectedResponse) {
if isAdd { if isAdd {
if filers.leaders.addLeaderIfVacant(address) { if m.leaders.addLeaderIfVacant(address) {
// has added the address as one leader // has added the address as one leader
result = append(result, &master_pb.KeepConnectedResponse{ result = append(result, &master_pb.KeepConnectedResponse{
ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{ ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{
@ -266,7 +237,7 @@ func ensureFilerLeaders(filers *Filers, isAdd bool, filerGroup FilerGroup, nodeT
}) })
} }
} else { } else {
if filers.leaders.removeLeaderIfExists(address) { if m.leaders.removeLeaderIfExists(address) {
result = append(result, &master_pb.KeepConnectedResponse{ result = append(result, &master_pb.KeepConnectedResponse{
ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{ ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{
@ -282,8 +253,8 @@ func ensureFilerLeaders(filers *Filers, isAdd bool, filerGroup FilerGroup, nodeT
var shortestDuration int64 = math.MaxInt64 var shortestDuration int64 = math.MaxInt64
now := time.Now() now := time.Now()
var candidateAddress pb.ServerAddress var candidateAddress pb.ServerAddress
for _, node := range filers.members { for _, node := range m.members {
if filers.leaders.isOneLeader(node.Address) { if m.leaders.isOneLeader(node.Address) {
continue continue
} }
duration := now.Sub(node.CreatedTs).Nanoseconds() duration := now.Sub(node.CreatedTs).Nanoseconds()
@ -293,7 +264,7 @@ func ensureFilerLeaders(filers *Filers, isAdd bool, filerGroup FilerGroup, nodeT
} }
} }
if candidateAddress != "" { if candidateAddress != "" {
filers.leaders.addLeaderIfVacant(candidateAddress) m.leaders.addLeaderIfVacant(candidateAddress)
// added a new leader // added a new leader
result = append(result, &master_pb.KeepConnectedResponse{ result = append(result, &master_pb.KeepConnectedResponse{
ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{ ClusterNodeUpdate: &master_pb.ClusterNodeUpdate{

View file

@ -16,7 +16,7 @@ func TestClusterAddRemoveNodes(t *testing.T) {
assert.Equal(t, []pb.ServerAddress{ assert.Equal(t, []pb.ServerAddress{
pb.ServerAddress("111:1"), pb.ServerAddress("111:1"),
pb.ServerAddress("111:2"), pb.ServerAddress("111:2"),
}, c.getFilers("", false).leaders.GetLeaders()) }, c.getGroupMembers("", "filer", true).leaders.GetLeaders())
c.AddClusterNode("", "filer", "", "", pb.ServerAddress("111:3"), "23.45") c.AddClusterNode("", "filer", "", "", pb.ServerAddress("111:3"), "23.45")
c.AddClusterNode("", "filer", "", "", pb.ServerAddress("111:4"), "23.45") c.AddClusterNode("", "filer", "", "", pb.ServerAddress("111:4"), "23.45")
@ -24,27 +24,27 @@ func TestClusterAddRemoveNodes(t *testing.T) {
pb.ServerAddress("111:1"), pb.ServerAddress("111:1"),
pb.ServerAddress("111:2"), pb.ServerAddress("111:2"),
pb.ServerAddress("111:3"), pb.ServerAddress("111:3"),
}, c.getFilers("", false).leaders.GetLeaders()) }, c.getGroupMembers("", "filer", true).leaders.GetLeaders())
c.AddClusterNode("", "filer", "", "", pb.ServerAddress("111:5"), "23.45") c.AddClusterNode("", "filer", "", "", pb.ServerAddress("111:5"), "23.45")
c.AddClusterNode("", "filer", "", "", pb.ServerAddress("111:6"), "23.45") c.AddClusterNode("", "filer", "", "", pb.ServerAddress("111:6"), "23.45")
c.RemoveClusterNode("", "filer", "", "", pb.ServerAddress("111:4")) c.RemoveClusterNode("", "filer", pb.ServerAddress("111:4"))
assert.Equal(t, []pb.ServerAddress{ assert.Equal(t, []pb.ServerAddress{
pb.ServerAddress("111:1"), pb.ServerAddress("111:1"),
pb.ServerAddress("111:2"), pb.ServerAddress("111:2"),
pb.ServerAddress("111:3"), pb.ServerAddress("111:3"),
}, c.getFilers("", false).leaders.GetLeaders()) }, c.getGroupMembers("", "filer", true).leaders.GetLeaders())
// remove oldest // remove oldest
c.RemoveClusterNode("", "filer", "", "", pb.ServerAddress("111:1")) c.RemoveClusterNode("", "filer", pb.ServerAddress("111:1"))
assert.Equal(t, []pb.ServerAddress{ assert.Equal(t, []pb.ServerAddress{
pb.ServerAddress("111:6"), pb.ServerAddress("111:6"),
pb.ServerAddress("111:2"), pb.ServerAddress("111:2"),
pb.ServerAddress("111:3"), pb.ServerAddress("111:3"),
}, c.getFilers("", false).leaders.GetLeaders()) }, c.getGroupMembers("", "filer", true).leaders.GetLeaders())
// remove oldest // remove oldest
c.RemoveClusterNode("", "filer", "", "", pb.ServerAddress("111:1")) c.RemoveClusterNode("", "filer", pb.ServerAddress("111:1"))
} }
@ -66,7 +66,7 @@ func TestConcurrentAddRemoveNodes(t *testing.T) {
go func(i int) { go func(i int) {
defer wg.Done() defer wg.Done()
address := strconv.Itoa(i) address := strconv.Itoa(i)
node := c.RemoveClusterNode("", "filer", "", "", pb.ServerAddress(address)) node := c.RemoveClusterNode("", "filer", pb.ServerAddress(address))
if len(node) == 0 { if len(node) == 0 {
t.Errorf("TestConcurrentAddRemoveNodes: node[%s] not found", address) t.Errorf("TestConcurrentAddRemoveNodes: node[%s] not found", address)

View file

@ -259,7 +259,7 @@ func (ms *MasterServer) KeepConnected(stream master_pb.Seaweed_KeepConnectedServ
} }
defer func() { defer func() {
for _, update := range ms.Cluster.RemoveClusterNode(req.FilerGroup, req.ClientType, cluster.DataCenter(req.DataCenter), cluster.Rack(req.Rack), peerAddress) { for _, update := range ms.Cluster.RemoveClusterNode(req.FilerGroup, req.ClientType, peerAddress) {
ms.broadcastToClients(update) ms.broadcastToClients(update)
} }
ms.deleteClient(clientName) ms.deleteClient(clientName)

View file

@ -10,14 +10,14 @@ import (
func (ms *MasterServer) ListClusterNodes(ctx context.Context, req *master_pb.ListClusterNodesRequest) (*master_pb.ListClusterNodesResponse, error) { func (ms *MasterServer) ListClusterNodes(ctx context.Context, req *master_pb.ListClusterNodesRequest) (*master_pb.ListClusterNodesResponse, error) {
resp := &master_pb.ListClusterNodesResponse{} resp := &master_pb.ListClusterNodesResponse{}
filerGroup := cluster.FilerGroup(req.FilerGroup) filerGroup := cluster.FilerGroupName(req.FilerGroup)
clusterNodes := ms.Cluster.ListClusterNode(filerGroup, req.ClientType) clusterNodes := ms.Cluster.ListClusterNode(filerGroup, req.ClientType)
for _, node := range clusterNodes { for _, node := range clusterNodes {
resp.ClusterNodes = append(resp.ClusterNodes, &master_pb.ListClusterNodesResponse_ClusterNode{ resp.ClusterNodes = append(resp.ClusterNodes, &master_pb.ListClusterNodesResponse_ClusterNode{
Address: string(node.Address), Address: string(node.Address),
Version: node.Version, Version: node.Version,
IsLeader: ms.Cluster.IsOneLeader(filerGroup, node.Address), IsLeader: ms.Cluster.IsOneLeader(filerGroup, req.ClientType, node.Address),
CreatedAtNs: node.CreatedTs.UnixNano(), CreatedAtNs: node.CreatedTs.UnixNano(),
DataCenter: string(node.DataCenter), DataCenter: string(node.DataCenter),
Rack: string(node.Rack), Rack: string(node.Rack),
@ -26,13 +26,13 @@ func (ms *MasterServer) ListClusterNodes(ctx context.Context, req *master_pb.Lis
return resp, nil return resp, nil
} }
func (ms *MasterServer) GetOneFiler(filerGroup cluster.FilerGroup) pb.ServerAddress { func (ms *MasterServer) GetOneFiler(filerGroup cluster.FilerGroupName) pb.ServerAddress {
clusterNodes := ms.Cluster.ListClusterNode(filerGroup, cluster.FilerType) clusterNodes := ms.Cluster.ListClusterNode(filerGroup, cluster.FilerType)
var filers []pb.ServerAddress var filers []pb.ServerAddress
for _, node := range clusterNodes { for _, node := range clusterNodes {
if ms.Cluster.IsOneLeader(filerGroup, node.Address) { if ms.Cluster.IsOneLeader(filerGroup, cluster.FilerType, node.Address) {
filers = append(filers, node.Address) filers = append(filers, node.Address)
} }
} }

View file

@ -287,7 +287,7 @@ func (ms *MasterServer) startAdminScripts() {
for { for {
time.Sleep(time.Duration(sleepMinutes) * time.Minute) time.Sleep(time.Duration(sleepMinutes) * time.Minute)
if ms.Topo.IsLeader() { if ms.Topo.IsLeader() {
shellOptions.FilerAddress = ms.GetOneFiler(cluster.FilerGroup(*shellOptions.FilerGroup)) shellOptions.FilerAddress = ms.GetOneFiler(cluster.FilerGroupName(*shellOptions.FilerGroup))
if shellOptions.FilerAddress == "" { if shellOptions.FilerAddress == "" {
continue continue
} }