1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520
|
package rethinkdb
import (
"errors"
"fmt"
"sort"
"strings"
"sync"
"sync/atomic"
"time"
"github.com/hailocab/go-hostpool"
"github.com/sirupsen/logrus"
"golang.org/x/net/context"
"gopkg.in/cenkalti/backoff.v2"
)
var errClusterClosed = errors.New("rethinkdb: cluster is closed")
const (
clusterWorking = 0
clusterClosed = 1
)
// A Cluster represents a connection to a RethinkDB cluster, a cluster is created
// by the Session and should rarely be created manually.
//
// The cluster keeps track of all nodes in the cluster and if requested can listen
// for cluster changes and start tracking a new node if one appears. Currently
// nodes are removed from the pool if they become unhealthy (100 failed queries).
// This should hopefully soon be replaced by a backoff system.
type Cluster struct {
opts *ConnectOpts
mu sync.RWMutex
seeds []Host // Initial host nodes specified by user.
hp hostpool.HostPool
nodes map[string]*Node // Active nodes in cluster.
closed int32 // 0 - working, 1 - closed
connFactory connFactory
discoverInterval time.Duration
}
// NewCluster creates a new cluster by connecting to the given hosts.
func NewCluster(hosts []Host, opts *ConnectOpts) (*Cluster, error) {
c := &Cluster{
hp: newHostPool(opts),
seeds: hosts,
opts: opts,
closed: clusterWorking,
connFactory: NewConnection,
}
err := c.run()
if err != nil {
return nil, err
}
return c, nil
}
func newHostPool(opts *ConnectOpts) hostpool.HostPool {
return hostpool.NewEpsilonGreedy([]string{}, opts.HostDecayDuration, &hostpool.LinearEpsilonValueCalculator{})
}
func (c *Cluster) run() error {
// Attempt to connect to each host and discover any additional hosts if host
// discovery is enabled
if err := c.connectCluster(); err != nil {
return err
}
if !c.IsConnected() {
return ErrNoConnectionsStarted
}
return nil
}
// Query executes a ReQL query using the cluster to connect to the database
func (c *Cluster) Query(ctx context.Context, q Query) (cursor *Cursor, err error) {
for i := 0; i < c.numRetries(); i++ {
var node *Node
var hpr hostpool.HostPoolResponse
node, hpr, err = c.GetNextNode()
if err != nil {
return nil, err
}
cursor, err = node.Query(ctx, q)
hpr.Mark(err)
if !shouldRetryQuery(q, err) {
break
}
}
return cursor, err
}
// Exec executes a ReQL query using the cluster to connect to the database
func (c *Cluster) Exec(ctx context.Context, q Query) (err error) {
for i := 0; i < c.numRetries(); i++ {
var node *Node
var hpr hostpool.HostPoolResponse
node, hpr, err = c.GetNextNode()
if err != nil {
return err
}
err = node.Exec(ctx, q)
hpr.Mark(err)
if !shouldRetryQuery(q, err) {
break
}
}
return err
}
// Server returns the server name and server UUID being used by a connection.
func (c *Cluster) Server() (response ServerResponse, err error) {
for i := 0; i < c.numRetries(); i++ {
var node *Node
var hpr hostpool.HostPoolResponse
node, hpr, err = c.GetNextNode()
if err != nil {
return ServerResponse{}, err
}
response, err = node.Server()
hpr.Mark(err)
// This query should not fail so retry if any error is detected
if err == nil {
break
}
}
return response, err
}
// SetInitialPoolCap sets the initial capacity of the connection pool.
func (c *Cluster) SetInitialPoolCap(n int) {
for _, node := range c.GetNodes() {
node.SetInitialPoolCap(n)
}
}
// SetMaxIdleConns sets the maximum number of connections in the idle
// connection pool.
func (c *Cluster) SetMaxIdleConns(n int) {
for _, node := range c.GetNodes() {
node.SetMaxIdleConns(n)
}
}
// SetMaxOpenConns sets the maximum number of open connections to the database.
func (c *Cluster) SetMaxOpenConns(n int) {
for _, node := range c.GetNodes() {
node.SetMaxOpenConns(n)
}
}
// Close closes the cluster
func (c *Cluster) Close(optArgs ...CloseOpts) error {
if c.isClosed() {
return nil
}
for _, node := range c.GetNodes() {
err := node.Close(optArgs...)
if err != nil {
return err
}
}
c.hp.Close()
atomic.StoreInt32(&c.closed, clusterClosed)
return nil
}
func (c *Cluster) isClosed() bool {
return atomic.LoadInt32(&c.closed) == clusterClosed
}
// discover attempts to find new nodes in the cluster using the current nodes
func (c *Cluster) discover() {
// Keep retrying with exponential backoff.
b := backoff.NewExponentialBackOff()
// Never finish retrying (max interval is still 60s)
b.MaxElapsedTime = 0
if c.discoverInterval != 0 {
b.InitialInterval = c.discoverInterval
}
// Keep trying to discover new nodes
for {
if c.isClosed() {
return
}
_ = backoff.RetryNotify(func() error {
if c.isClosed() {
return backoff.Permanent(errClusterClosed)
}
// If no hosts try seeding nodes
if len(c.GetNodes()) == 0 {
return c.connectCluster()
}
return c.listenForNodeChanges()
}, b, func(err error, wait time.Duration) {
Log.Debugf("Error discovering hosts %s, waiting: %s", err, wait)
})
}
}
// listenForNodeChanges listens for changes to node status using change feeds.
// This function will block until the query fails
func (c *Cluster) listenForNodeChanges() error {
// Start listening to changes from a random active node
node, hpr, err := c.GetNextNode()
if err != nil {
return err
}
q, err := newQuery(
DB(SystemDatabase).Table(ServerStatusSystemTable).Changes(ChangesOpts{IncludeInitial: true}),
map[string]interface{}{},
c.opts,
)
if err != nil {
return fmt.Errorf("Error building query: %s", err)
}
cursor, err := node.Query(context.Background(), q) // no need for timeout due to Changes()
if err != nil {
hpr.Mark(err)
return err
}
defer func() { _ = cursor.Close() }()
// Keep reading node status updates from changefeed
var result struct {
NewVal *nodeStatus `rethinkdb:"new_val"`
OldVal *nodeStatus `rethinkdb:"old_val"`
}
for cursor.Next(&result) {
addr := fmt.Sprintf("%s:%d", result.NewVal.Network.Hostname, result.NewVal.Network.ReqlPort)
addr = strings.ToLower(addr)
if result.NewVal != nil && result.OldVal == nil {
// added new node
if !c.nodeExists(result.NewVal.ID) {
// Connect to node using exponential backoff (give up after waiting 5s)
// to give the node time to start-up.
b := backoff.NewExponentialBackOff()
b.MaxElapsedTime = time.Second * 5
err = backoff.Retry(func() error {
node, err := c.connectNodeWithStatus(result.NewVal)
if err == nil {
c.addNode(node)
Log.WithFields(logrus.Fields{
"id": node.ID,
"host": node.Host.String(),
}).Debug("Connected to node")
}
return err
}, b)
if err != nil {
return err
}
}
} else if result.OldVal != nil && result.NewVal == nil {
// removed old node
oldNode := c.removeNode(result.OldVal.ID)
if oldNode != nil {
_ = oldNode.Close()
}
} else {
// node updated
// nothing to do - assuming node can't change it's hostname in a single Changes() message
}
}
err = cursor.Err()
hpr.Mark(err)
return err
}
func (c *Cluster) connectCluster() error {
nodeSet := map[string]*Node{}
var attemptErr error
// Attempt to connect to each seed host
for _, host := range c.seeds {
conn, err := c.connFactory(host.String(), c.opts)
if err != nil {
attemptErr = err
Log.Warnf("Error creating connection: %s", err.Error())
continue
}
svrRsp, err := conn.Server()
if err != nil {
attemptErr = err
Log.Warnf("Error fetching server ID: %s", err)
_ = conn.Close()
continue
}
_ = conn.Close()
node, err := c.connectNode(svrRsp.ID, []Host{host})
if err != nil {
attemptErr = err
Log.Warnf("Error connecting to node: %s", err)
continue
}
if _, ok := nodeSet[node.ID]; !ok {
Log.WithFields(logrus.Fields{
"id": node.ID,
"host": node.Host.String(),
}).Debug("Connected to node")
nodeSet[node.ID] = node
} else {
// dublicate node
_ = node.Close()
}
}
// If no nodes were contactable then return the last error, this does not
// include driver errors such as if there was an issue building the
// query
if len(nodeSet) == 0 {
if attemptErr != nil {
return attemptErr
}
return ErrNoConnections
}
var nodes []*Node
for _, node := range nodeSet {
nodes = append(nodes, node)
}
c.replaceNodes(nodes)
if c.opts.DiscoverHosts {
go c.discover()
}
return nil
}
func (c *Cluster) connectNodeWithStatus(s *nodeStatus) (*Node, error) {
aliases := make([]Host, len(s.Network.CanonicalAddresses))
for i, aliasAddress := range s.Network.CanonicalAddresses {
aliases[i] = NewHost(aliasAddress.Host, int(s.Network.ReqlPort))
}
return c.connectNode(s.ID, aliases)
}
func (c *Cluster) connectNode(id string, aliases []Host) (*Node, error) {
var pool *Pool
var err error
for len(aliases) > 0 {
pool, err = newPool(aliases[0], c.opts, c.connFactory)
if err != nil {
aliases = aliases[1:]
continue
}
err = pool.Ping()
if err != nil {
aliases = aliases[1:]
continue
}
// Ping successful so break out of loop
break
}
if err != nil {
return nil, err
}
if len(aliases) == 0 {
return nil, ErrInvalidNode
}
return newNode(id, aliases, pool), nil
}
// IsConnected returns true if cluster has nodes and is not already connClosed.
func (c *Cluster) IsConnected() bool {
return (len(c.GetNodes()) > 0) && !c.isClosed()
}
// GetNextNode returns a random node on the cluster
func (c *Cluster) GetNextNode() (*Node, hostpool.HostPoolResponse, error) {
if !c.IsConnected() {
return nil, nil, ErrNoConnections
}
c.mu.RLock()
defer c.mu.RUnlock()
nodes := c.nodes
hpr := c.hp.Get()
if n, ok := nodes[hpr.Host()]; ok {
if !n.Closed() {
return n, hpr, nil
}
}
return nil, nil, ErrNoConnections
}
// GetNodes returns a list of all nodes in the cluster
func (c *Cluster) GetNodes() []*Node {
c.mu.RLock()
defer c.mu.RUnlock()
nodes := make([]*Node, 0, len(c.nodes))
for _, n := range c.nodes {
nodes = append(nodes, n)
}
return nodes
}
func (c *Cluster) nodeExists(nodeID string) bool {
c.mu.RLock()
defer c.mu.RUnlock()
for _, node := range c.nodes {
if node.ID == nodeID {
return true
}
}
return false
}
func (c *Cluster) addNode(node *Node) {
host := node.Host.String()
c.mu.Lock()
defer c.mu.Unlock()
if _, exist := c.nodes[host]; exist {
// addNode() should be called only if the node doesn't exist
return
}
c.nodes[host] = node
hosts := make([]string, 0, len(c.nodes))
for _, n := range c.nodes {
hosts = append(hosts, n.Host.String())
}
c.hp.SetHosts(hosts)
}
func (c *Cluster) replaceNodes(nodes []*Node) {
nodesMap := make(map[string]*Node, len(nodes))
hosts := make([]string, len(nodes))
for i, node := range nodes {
host := node.Host.String()
nodesMap[host] = node
hosts[i] = host
}
sort.Strings(hosts) // unit tests stability
c.mu.Lock()
c.nodes = nodesMap
c.hp.SetHosts(hosts)
c.mu.Unlock()
}
func (c *Cluster) removeNode(nodeID string) *Node {
c.mu.Lock()
defer c.mu.Unlock()
var rmNode *Node
for _, node := range c.nodes {
if node.ID == nodeID {
rmNode = node
break
}
}
if rmNode == nil {
return nil
}
delete(c.nodes, rmNode.Host.String())
hosts := make([]string, 0, len(c.nodes))
for _, n := range c.nodes {
hosts = append(hosts, n.Host.String())
}
c.hp.SetHosts(hosts)
return rmNode
}
func (c *Cluster) numRetries() int {
if n := c.opts.NumRetries; n > 0 {
return n
}
return 3
}
|