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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565
|
// Copyright 2013-2018 The NATS Authors
// 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,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package server
import (
"bytes"
"crypto/tls"
"encoding/json"
"fmt"
"math/rand"
"net"
"net/url"
"runtime"
"strconv"
"strings"
"sync/atomic"
"time"
)
// RouteType designates the router type
type RouteType int
// Type of Route
const (
// This route we learned from speaking to other routes.
Implicit RouteType = iota
// This route was explicitly configured.
Explicit
)
const (
// RouteProtoZero is the original Route protocol from 2009.
// http://nats.io/documentation/internals/nats-protocol/
RouteProtoZero = iota
// RouteProtoInfo signals a route can receive more then the original INFO block.
// This can be used to update remote cluster permissions, etc...
RouteProtoInfo
// RouteProtoV2 is the new route/cluster protocol that provides account support.
RouteProtoV2
)
// Used by tests
var testRouteProto = RouteProtoV2
type route struct {
remoteID string
didSolicit bool
retry bool
routeType RouteType
url *url.URL
authRequired bool
tlsRequired bool
closed bool
connectURLs []string
replySubs map[*subscription]*time.Timer
}
type connectInfo struct {
Echo bool `json:"echo"`
Verbose bool `json:"verbose"`
Pedantic bool `json:"pedantic"`
User string `json:"user,omitempty"`
Pass string `json:"pass,omitempty"`
TLS bool `json:"tls_required"`
Name string `json:"name"`
}
// Route protocol constants
const (
ConProto = "CONNECT %s" + _CRLF_
InfoProto = "INFO %s" + _CRLF_
)
// This will add a timer to watch over remote reply subjects in case
// they fail to receive a response. The duration will be taken from the
// accounts map timeout to match.
// Lock should be held upon entering.
func (c *client) addReplySubTimeout(acc *Account, sub *subscription, d time.Duration) {
if c.route.replySubs == nil {
c.route.replySubs = make(map[*subscription]*time.Timer)
}
rs := c.route.replySubs
rs[sub] = time.AfterFunc(d, func() {
c.mu.Lock()
delete(rs, sub)
sub.max = 0
c.mu.Unlock()
c.unsubscribe(acc, sub, true)
})
}
// removeReplySub is called when we trip the max on remoteReply subs.
func (c *client) removeReplySub(sub *subscription) {
if sub == nil {
return
}
// Lookup the account based on sub.sid.
if i := bytes.Index(sub.sid, []byte(" ")); i > 0 {
// First part of SID for route is account name.
acc := c.srv.LookupAccount(string(sub.sid[:i]))
acc.sl.Remove(sub)
c.mu.Lock()
c.removeReplySubTimeout(sub)
c.mu.Unlock()
}
}
// removeReplySubTimeout will remove a timer if it exists.
// Lock should be held upon entering.
func (c *client) removeReplySubTimeout(sub *subscription) {
// Remove any reply sub timer if it exists.
if c.route.replySubs == nil {
return
}
if t, ok := c.route.replySubs[sub]; ok {
t.Stop()
delete(c.route.replySubs, sub)
}
}
func (c *client) processAccountSub(arg []byte) error {
// Placeholder in case we add in to the protocol active senders of
// informtation. For now we do not do account interest propagation.
c.traceInOp("A+", arg)
return nil
}
func (c *client) processAccountUnsub(arg []byte) {
// Placeholder in case we add in to the protocol active senders of
// informtation. For now we do not do account interest propagation.
c.traceInOp("A-", arg)
}
// Process an inbound RMSG specification from the remote route.
func (c *client) processRoutedMsgArgs(trace bool, arg []byte) error {
if trace {
c.traceInOp("RMSG", arg)
}
// Unroll splitArgs to avoid runtime/heap issues
a := [MAX_MSG_ARGS][]byte{}
args := a[:0]
start := -1
for i, b := range arg {
switch b {
case ' ', '\t', '\r', '\n':
if start >= 0 {
args = append(args, arg[start:i])
start = -1
}
default:
if start < 0 {
start = i
}
}
}
if start >= 0 {
args = append(args, arg[start:])
}
c.pa.arg = arg
switch len(args) {
case 0, 1, 2:
return fmt.Errorf("processRoutedMsgArgs Parse Error: '%s'", args)
case 3:
c.pa.reply = nil
c.pa.queues = nil
c.pa.szb = args[2]
c.pa.size = parseSize(args[2])
case 4:
c.pa.reply = args[2]
c.pa.queues = nil
c.pa.szb = args[3]
c.pa.size = parseSize(args[3])
default:
// args[2] is our reply indicator. Should be + or | normally.
if len(args[2]) != 1 {
return fmt.Errorf("processRoutedMsgArgs Bad or Missing Reply Indicator: '%s'", args[2])
}
switch args[2][0] {
case '+':
c.pa.reply = args[3]
case '|':
c.pa.reply = nil
default:
return fmt.Errorf("processRoutedMsgArgs Bad or Missing Reply Indicator: '%s'", args[2])
}
// Grab size.
c.pa.szb = args[len(args)-1]
c.pa.size = parseSize(c.pa.szb)
// Grab queue names.
if c.pa.reply != nil {
c.pa.queues = args[4 : len(args)-1]
} else {
c.pa.queues = args[3 : len(args)-1]
}
}
if c.pa.size < 0 {
return fmt.Errorf("processRoutedMsgArgs Bad or Missing Size: '%s'", args)
}
// Common ones processed after check for arg length
c.pa.account = args[0]
c.pa.subject = args[1]
c.pa.rcache = arg[:len(args[0])+len(args[1])+1]
return nil
}
const (
maxRouteCacheSize = 32768
pruneRouteCacheSize = 512
)
// routeCache is for L1 semantics for inbound messages from a route to mimic the performance of clients.
type routeCache struct {
acc *Account
results *SublistResult
genid uint64
}
// pruneRouteCache will prune off a random number of cache entries.
func (c *client) pruneRouteCache() {
n := 0
for cacheKey := range c.in.rcache {
delete(c.in.rcache, cacheKey)
if n++; n > pruneRouteCacheSize {
break
}
}
}
// processInboundRouteMsg is called to process an inbound msg from a route.
func (c *client) processInboundRoutedMsg(msg []byte) {
// Update statistics
c.in.msgs++
// The msg includes the CR_LF, so pull back out for accounting.
c.in.bytes += len(msg) - LEN_CR_LF
if c.trace {
c.traceMsg(msg)
}
if c.opts.Verbose {
c.sendOK()
}
// Mostly under testing scenarios.
if c.srv == nil {
return
}
var (
acc *Account
rc *routeCache
r *SublistResult
ok bool
)
// Check our cache first.
if rc, ok = c.in.rcache[string(c.pa.rcache)]; ok {
// Check the genid to see if it's still valid.
if genid := atomic.LoadUint64(&rc.acc.sl.genid); genid != rc.genid {
ok = false
delete(c.in.rcache, string(c.pa.rcache))
} else {
acc = rc.acc
r = rc.results
}
}
if !ok {
// Match correct account and sublist.
acc = c.srv.LookupAccount(string(c.pa.account))
if acc == nil {
c.Debugf("Unknown account %q for routed message on subject: %q", c.pa.account, c.pa.subject)
return
}
// Match against the account sublist.
r = acc.sl.Match(string(c.pa.subject))
// Store in our cache
c.in.rcache[string(c.pa.rcache)] = &routeCache{acc, r, atomic.LoadUint64(&acc.sl.genid)}
// Check if we need to prune.
if len(c.in.rcache) > maxRouteCacheSize {
c.pruneRouteCache()
}
}
// Check to see if we need to map/route to another account.
if acc.imports.services != nil {
c.checkForImportServices(acc, msg)
}
// Check for no interest, short circuit if so.
// This is the fanout scale.
if len(r.psubs)+len(r.qsubs) == 0 {
return
}
// Check to see if we have a routed message with a service reply.
if isServiceReply(c.pa.reply) && acc != nil {
// Need to add a sub here for local interest to send a response back
// to the originating server/requestor where it will be re-mapped.
sid := make([]byte, 0, len(acc.Name)+len(c.pa.reply)+1)
sid = append(sid, acc.Name...)
sid = append(sid, ' ')
sid = append(sid, c.pa.reply...)
// Copy off the reply since otherwise we are referencing a buffer that will be reused.
reply := make([]byte, len(c.pa.reply))
copy(reply, c.pa.reply)
sub := &subscription{client: c, subject: reply, sid: sid, max: 1}
if err := acc.sl.Insert(sub); err != nil {
c.Errorf("Could not insert subscription: %v", err)
} else {
ttl := acc.AutoExpireTTL()
c.mu.Lock()
c.subs[string(sid)] = sub
c.addReplySubTimeout(acc, sub, ttl)
c.mu.Unlock()
}
}
c.processMsgResults(acc, r, msg, c.pa.subject, c.pa.reply)
}
// Lock should be held entering here.
func (c *client) sendConnect(tlsRequired bool) {
var user, pass string
if userInfo := c.route.url.User; userInfo != nil {
user = userInfo.Username()
pass, _ = userInfo.Password()
}
cinfo := connectInfo{
Echo: true,
Verbose: false,
Pedantic: false,
User: user,
Pass: pass,
TLS: tlsRequired,
Name: c.srv.info.ID,
}
b, err := json.Marshal(cinfo)
if err != nil {
c.Errorf("Error marshaling CONNECT to route: %v\n", err)
c.closeConnection(ProtocolViolation)
return
}
c.sendProto([]byte(fmt.Sprintf(ConProto, b)), true)
}
// Process the info message if we are a route.
func (c *client) processRouteInfo(info *Info) {
c.mu.Lock()
// Connection can be closed at any time (by auth timeout, etc).
// Does not make sense to continue here if connection is gone.
if c.route == nil || c.nc == nil {
c.mu.Unlock()
return
}
s := c.srv
remoteID := c.route.remoteID
// We receive an INFO from a server that informs us about another server,
// so the info.ID in the INFO protocol does not match the ID of this route.
if remoteID != "" && remoteID != info.ID {
c.mu.Unlock()
// Process this implicit route. We will check that it is not an explicit
// route and/or that it has not been connected already.
s.processImplicitRoute(info)
return
}
// Need to set this for the detection of the route to self to work
// in closeConnection().
c.route.remoteID = info.ID
// Get the route's proto version
c.opts.Protocol = info.Proto
// Detect route to self.
if c.route.remoteID == s.info.ID {
c.mu.Unlock()
c.closeConnection(DuplicateRoute)
return
}
// Copy over important information.
c.route.authRequired = info.AuthRequired
c.route.tlsRequired = info.TLSRequired
// If this is an update due to config reload on the remote server,
// need to possibly send local subs to the remote server.
if c.flags.isSet(infoReceived) {
s.updateRemoteRoutePerms(c, info)
c.mu.Unlock()
return
}
// Copy over permissions as well.
c.opts.Import = info.Import
c.opts.Export = info.Export
// If we do not know this route's URL, construct one on the fly
// from the information provided.
if c.route.url == nil {
// Add in the URL from host and port
hp := net.JoinHostPort(info.Host, strconv.Itoa(info.Port))
url, err := url.Parse(fmt.Sprintf("nats-route://%s/", hp))
if err != nil {
c.Errorf("Error parsing URL from INFO: %v\n", err)
c.mu.Unlock()
c.closeConnection(ParseError)
return
}
c.route.url = url
}
// Mark that the INFO protocol has been received. Will allow
// to detect INFO updates.
c.flags.set(infoReceived)
// Check to see if we have this remote already registered.
// This can happen when both servers have routes to each other.
c.mu.Unlock()
if added, sendInfo := s.addRoute(c, info); added {
c.Debugf("Registering remote route %q", info.ID)
// Send our subs to the other side.
s.sendSubsToRoute(c)
// sendInfo will be false if the route that we just accepted
// is the only route there is.
if sendInfo {
// The incoming INFO from the route will have IP set
// if it has Cluster.Advertise. In that case, use that
// otherwise contruct it from the remote TCP address.
if info.IP == "" {
// Need to get the remote IP address.
c.mu.Lock()
switch conn := c.nc.(type) {
case *net.TCPConn, *tls.Conn:
addr := conn.RemoteAddr().(*net.TCPAddr)
info.IP = fmt.Sprintf("nats-route://%s/", net.JoinHostPort(addr.IP.String(),
strconv.Itoa(info.Port)))
default:
info.IP = c.route.url.String()
}
c.mu.Unlock()
}
// Now let the known servers know about this new route
s.forwardNewRouteInfoToKnownServers(info)
}
// Unless disabled, possibly update the server's INFO protocol
// and send to clients that know how to handle async INFOs.
if !s.getOpts().Cluster.NoAdvertise {
s.addClientConnectURLsAndSendINFOToClients(info.ClientConnectURLs)
}
} else {
c.Debugf("Detected duplicate remote route %q", info.ID)
c.closeConnection(DuplicateRoute)
}
}
// Possibly sends local subscriptions interest to this route
// based on changes in the remote's Export permissions.
// Lock assumed held on entry
func (s *Server) updateRemoteRoutePerms(route *client, info *Info) {
// Interested only on Export permissions for the remote server.
// Create "fake" clients that we will use to check permissions
// using the old permissions...
oldPerms := &RoutePermissions{Export: route.opts.Export}
oldPermsTester := &client{}
oldPermsTester.setRoutePermissions(oldPerms)
// and the new ones.
newPerms := &RoutePermissions{Export: info.Export}
newPermsTester := &client{}
newPermsTester.setRoutePermissions(newPerms)
route.opts.Import = info.Import
route.opts.Export = info.Export
var (
_localSubs [4096]*subscription
localSubs = _localSubs[:0]
)
// FIXME(dlc) - Add account scoping.
s.gacc.sl.localSubs(&localSubs)
route.sendRouteSubProtos(localSubs, false, func(sub *subscription) bool {
subj := string(sub.subject)
// If the remote can now export but could not before, and this server can import this
// subject, then send SUB protocol.
if newPermsTester.canExport(subj) && !oldPermsTester.canExport(subj) && route.canImport(subj) {
return true
}
return false
})
}
// sendAsyncInfoToClients sends an INFO protocol to all
// connected clients that accept async INFO updates.
// The server lock is held on entry.
func (s *Server) sendAsyncInfoToClients() {
// If there are no clients supporting async INFO protocols, we are done.
// Also don't send if we are shutting down...
if s.cproto == 0 || s.shutdown {
return
}
for _, c := range s.clients {
c.mu.Lock()
// Here, we are going to send only to the clients that are fully
// registered (server has received CONNECT and first PING). For
// clients that are not at this stage, this will happen in the
// processing of the first PING (see client.processPing)
if c.opts.Protocol >= ClientProtoInfo && c.flags.isSet(firstPongSent) {
// sendInfo takes care of checking if the connection is still
// valid or not, so don't duplicate tests here.
c.sendInfo(c.generateClientInfoJSON(s.copyInfo()))
}
c.mu.Unlock()
}
}
// This will process implicit route information received from another server.
// We will check to see if we have configured or are already connected,
// and if so we will ignore. Otherwise we will attempt to connect.
func (s *Server) processImplicitRoute(info *Info) {
remoteID := info.ID
s.mu.Lock()
defer s.mu.Unlock()
// Don't connect to ourself
if remoteID == s.info.ID {
return
}
// Check if this route already exists
if _, exists := s.remotes[remoteID]; exists {
return
}
// Check if we have this route as a configured route
if s.hasThisRouteConfigured(info) {
return
}
// Initiate the connection, using info.IP instead of info.URL here...
r, err := url.Parse(info.IP)
if err != nil {
s.Errorf("Error parsing URL from INFO: %v\n", err)
return
}
// Snapshot server options.
opts := s.getOpts()
if info.AuthRequired {
r.User = url.UserPassword(opts.Cluster.Username, opts.Cluster.Password)
}
s.startGoRoutine(func() { s.connectToRoute(r, false) })
}
// hasThisRouteConfigured returns true if info.Host:info.Port is present
// in the server's opts.Routes, false otherwise.
// Server lock is assumed to be held by caller.
func (s *Server) hasThisRouteConfigured(info *Info) bool {
urlToCheckExplicit := strings.ToLower(net.JoinHostPort(info.Host, strconv.Itoa(info.Port)))
for _, ri := range s.getOpts().Routes {
if strings.ToLower(ri.Host) == urlToCheckExplicit {
return true
}
}
return false
}
// forwardNewRouteInfoToKnownServers sends the INFO protocol of the new route
// to all routes known by this server. In turn, each server will contact this
// new route.
func (s *Server) forwardNewRouteInfoToKnownServers(info *Info) {
s.mu.Lock()
defer s.mu.Unlock()
b, _ := json.Marshal(info)
infoJSON := []byte(fmt.Sprintf(InfoProto, b))
for _, r := range s.routes {
r.mu.Lock()
if r.route.remoteID != info.ID {
r.sendInfo(infoJSON)
}
r.mu.Unlock()
}
}
// canImport is whether or not we will send a SUB for interest to the other side.
// This is for ROUTER connections only.
// Lock is held on entry.
func (c *client) canImport(subject string) bool {
// Use pubAllowed() since this checks Publish permissions which
// is what Import maps to.
return c.pubAllowed(subject)
}
// canExport is whether or not we will accept a SUB from the remote for a given subject.
// This is for ROUTER connections only.
// Lock is held on entry
func (c *client) canExport(subject string) bool {
// Use canSubscribe() since this checks Subscribe permissions which
// is what Export maps to.
return c.canSubscribe(subject)
}
// Initialize or reset cluster's permissions.
// This is for ROUTER connections only.
// Client lock is held on entry
func (c *client) setRoutePermissions(perms *RoutePermissions) {
// Reset if some were set
if perms == nil {
c.perms = nil
c.mperms = nil
return
}
// Convert route permissions to user permissions.
// The Import permission is mapped to Publish
// and Export permission is mapped to Subscribe.
// For meaning of Import/Export, see canImport and canExport.
p := &Permissions{
Publish: perms.Import,
Subscribe: perms.Export,
}
c.setPermissions(p)
}
// Type used to hold a list of subs on a per account basis.
type asubs struct {
acc *Account
subs []*subscription
}
// removeRemoteSubs will walk the subs and remove them from the appropriate account.
func (c *client) removeRemoteSubs() {
// We need to gather these on a per account basis.
// FIXME(dlc) - We should be smarter about this..
as := map[string]*asubs{}
c.mu.Lock()
srv := c.srv
subs := c.subs
c.subs = make(map[string]*subscription)
c.mu.Unlock()
for key, sub := range subs {
c.mu.Lock()
sub.max = 0
c.mu.Unlock()
// Grab the account
accountName := strings.Fields(key)[0]
ase := as[accountName]
if ase == nil {
acc := srv.LookupAccount(accountName)
if acc == nil {
continue
}
as[accountName] = &asubs{acc: acc, subs: []*subscription{sub}}
} else {
ase.subs = append(ase.subs, sub)
}
}
// Now remove the subs by batch for each account sublist.
for _, ase := range as {
c.Debugf("Removing %d subscriptions for account %q", len(ase.subs), ase.acc.Name)
ase.acc.sl.RemoveBatch(ase.subs)
}
}
// Indicates no more interest in the given account/subject for the remote side.
func (c *client) processRemoteUnsub(arg []byte) (err error) {
c.traceInOp("RS-", arg)
// Indicate any activity, so pub and sub or unsubs.
c.in.subs++
srv := c.srv
if srv == nil {
return nil
}
args := splitArg(arg)
sub := &subscription{client: c}
switch len(args) {
case 2:
sub.queue = nil
case 3:
sub.queue = args[2]
default:
return fmt.Errorf("processRemoteUnSub Parse Error: '%s'", arg)
}
sub.subject = args[1]
// Lookup the account
accountName := string(args[0])
acc := c.srv.LookupAccount(accountName)
if acc == nil {
c.Debugf("Unknown account %q for subject %q", accountName, sub.subject)
// Mark this account as not interested since we received a RS- and we
// do not have any record of it.
return nil
}
c.mu.Lock()
if c.nc == nil {
c.mu.Unlock()
return nil
}
// We store local subs by account and subject and optionally queue name.
// RS- will have the arg exactly as the key.
key := string(arg)
if sub, ok := c.subs[key]; ok {
delete(c.subs, key)
acc.sl.Remove(sub)
c.removeReplySubTimeout(sub)
}
c.mu.Unlock()
if c.opts.Verbose {
c.sendOK()
}
return nil
}
func (c *client) processRemoteSub(argo []byte) (err error) {
c.traceInOp("RS+", argo)
// Indicate activity.
c.in.subs++
srv := c.srv
if srv == nil {
return nil
}
// Copy so we do not reference a potentially large buffer
arg := make([]byte, len(argo))
copy(arg, argo)
args := splitArg(arg)
sub := &subscription{client: c}
switch len(args) {
case 2:
sub.queue = nil
case 4:
sub.queue = args[2]
sub.qw = int32(parseSize(args[3]))
default:
return fmt.Errorf("processRemoteSub Parse Error: '%s'", arg)
}
sub.subject = args[1]
// Lookup the account
// FIXME(dlc) - This may start having lots of contention?
accountName := string(args[0])
acc := c.srv.LookupAccount(accountName)
if acc == nil {
if !srv.NewAccountsAllowed() {
c.Debugf("Unknown account %q for subject %q", accountName, sub.subject)
return nil
}
acc, _ = srv.LookupOrRegisterAccount(accountName)
}
c.mu.Lock()
if c.nc == nil {
c.mu.Unlock()
return nil
}
// Check permissions if applicable.
if !c.canExport(string(sub.subject)) {
c.mu.Unlock()
c.Debugf("Can not export %q, ignoring remote subscription request", sub.subject)
return nil
}
// Check if we have a maximum on the number of subscriptions.
if c.msubs > 0 && len(c.subs) >= c.msubs {
c.mu.Unlock()
c.maxSubsExceeded()
return nil
}
// We store local subs by account and subject and optionally queue name.
// If we have a queue it will have a trailing weight which we do not want.
if sub.queue != nil {
sub.sid = arg[:len(arg)-len(args[3])-1]
} else {
sub.sid = arg
}
key := string(sub.sid)
osub := c.subs[key]
if osub == nil {
c.subs[string(key)] = sub
// Now place into the account sl.
if err = acc.sl.Insert(sub); err != nil {
delete(c.subs, key)
c.mu.Unlock()
c.Errorf("Could not insert subscription: %v", err)
c.sendErr("Invalid Subscription")
return nil
}
} else if sub.queue != nil {
// For a queue we need to update the weight.
atomic.StoreInt32(&osub.qw, sub.qw)
acc.sl.UpdateRemoteQSub(osub)
}
c.mu.Unlock()
if c.opts.Verbose {
c.sendOK()
}
return nil
}
// sendSubsToRoute will send over our subject interest to
// the remote side. For each account we will send the
// complete interest for all subjects, both normal as a binary
// and queue group weights.
func (s *Server) sendSubsToRoute(route *client) {
// Send over our account subscriptions.
var _accs [4096]*Account
accs := _accs[:0]
// copy accounts into array first
s.mu.Lock()
for _, a := range s.accounts {
accs = append(accs, a)
}
s.mu.Unlock()
var raw [4096]*subscription
var closed bool
route.mu.Lock()
for _, a := range accs {
subs := raw[:0]
a.mu.RLock()
for key, rme := range a.rm {
// FIXME(dlc) - Just pass rme around.
// Construct a sub on the fly
var qn []byte
subEnd := len(key)
if qi := rme.qi; qi > 0 {
subEnd = int(qi) - 1
qn = []byte(key[qi:])
}
sub := &subscription{subject: []byte(key[:subEnd]), queue: qn, qw: rme.n}
subs = append(subs, sub)
}
a.mu.RUnlock()
closed = route.sendRouteSubProtos(subs, false, func(sub *subscription) bool {
return route.canImport(string(sub.subject))
})
if closed {
route.mu.Unlock()
return
}
}
route.mu.Unlock()
if !closed {
route.Debugf("Sent local subscriptions to route")
}
}
// Sends SUBs protocols for the given subscriptions. If a filter is specified, it is
// invoked for each subscription. If the filter returns false, the subscription is skipped.
// This function may release the route's lock due to flushing of outbound data. A boolean
// is returned to indicate if the connection has been closed during this call.
// Lock is held on entry.
func (c *client) sendRouteSubProtos(subs []*subscription, trace bool, filter func(sub *subscription) bool) bool {
return c.sendRouteSubOrUnSubProtos(subs, true, trace, filter)
}
// Sends UNSUBs protocols for the given subscriptions. If a filter is specified, it is
// invoked for each subscription. If the filter returns false, the subscription is skipped.
// This function may release the route's lock due to flushing of outbound data. A boolean
// is returned to indicate if the connection has been closed during this call.
// Lock is held on entry.
func (c *client) sendRouteUnSubProtos(subs []*subscription, trace bool, filter func(sub *subscription) bool) bool {
return c.sendRouteSubOrUnSubProtos(subs, false, trace, filter)
}
// Low-level function that sends SUBs or UNSUBs protcols for the given subscriptions.
// Use sendRouteSubProtos or sendRouteUnSubProtos instead for clarity.
// Lock is held on entry.
func (c *client) sendRouteSubOrUnSubProtos(subs []*subscription, isSubProto, trace bool, filter func(sub *subscription) bool) bool {
const staticBufSize = maxBufSize * 2
var (
_buf [staticBufSize]byte // array on stack
buf = _buf[:0] // our buffer will initially point to the stack buffer
mbs = staticBufSize // max size of the buffer
mpMax = int(c.out.mp / 2) // 50% of max_pending
closed bool
)
// We need to make sure that we stay below the user defined max pending bytes.
if mbs > mpMax {
mbs = mpMax
}
for _, sub := range subs {
if filter != nil && !filter(sub) {
continue
}
// Determine the account. If sub has an ImportMap entry, use that, otherwise scoped to
// client. Default to global if all else fails.
var accName string
if sub.im != nil {
accName = sub.im.acc.Name
} else if sub.client != nil && sub.client.acc != nil {
accName = sub.client.acc.Name
} else {
accName = globalAccountName
}
// Check if proto is going to fit.
curSize := len(buf)
// "RS+/- " + account + " " + subject + " " [+ queue + " " + weight] + CRLF
curSize += 4 + len(accName) + 1 + len(sub.subject) + 1 + 2
if len(sub.queue) > 0 {
curSize += len(sub.queue)
if isSubProto {
// Estimate weightlen in 1000s
curSize += 1 + 4
}
}
if curSize >= mbs {
if c.queueOutbound(buf) {
// Need to allocate new array
buf = make([]byte, 0, mbs)
} else {
// We can reuse previous buffer
buf = buf[:0]
}
// Update last activity because flushOutbound() will release
// the lock, which could cause pingTimer to think that this
// connection is stale otherwise.
c.last = time.Now()
c.flushOutbound()
if closed = c.flags.isSet(clearConnection); closed {
break
}
}
as := len(buf)
if isSubProto {
buf = append(buf, []byte("RS+ ")...)
} else {
buf = append(buf, []byte("RS- ")...)
}
buf = append(buf, []byte(accName)...)
buf = append(buf, ' ')
buf = append(buf, sub.subject...)
if len(sub.queue) > 0 {
buf = append(buf, ' ')
buf = append(buf, sub.queue...)
// Send our weight if we are a sub proto
if isSubProto {
buf = append(buf, ' ')
var b [12]byte
var i = len(b)
for l := sub.qw; l > 0; l /= 10 {
i--
b[i] = digits[l%10]
}
buf = append(buf, b[i:]...)
}
}
if trace {
c.traceOutOp("", buf[as:])
}
buf = append(buf, []byte(CR_LF)...)
}
if !closed && len(buf) > 0 {
c.queueOutbound(buf)
c.flushOutbound()
closed = c.flags.isSet(clearConnection)
}
return closed
}
func (s *Server) createRoute(conn net.Conn, rURL *url.URL) *client {
// Snapshot server options.
opts := s.getOpts()
didSolicit := rURL != nil
r := &route{didSolicit: didSolicit}
for _, route := range opts.Routes {
if rURL != nil && (strings.ToLower(rURL.Host) == strings.ToLower(route.Host)) {
r.routeType = Explicit
}
}
c := &client{srv: s, nc: conn, opts: clientOpts{}, typ: ROUTER, route: r}
// Grab server variables
s.mu.Lock()
s.generateRouteInfoJSON()
infoJSON := s.routeInfoJSON
authRequired := s.routeInfo.AuthRequired
tlsRequired := s.routeInfo.TLSRequired
s.mu.Unlock()
// Grab lock
c.mu.Lock()
// Initialize
c.initClient()
// Initialize the route cache.
c.in.rcache = make(map[string]*routeCache, maxRouteCacheSize)
if didSolicit {
// Do this before the TLS code, otherwise, in case of failure
// and if route is explicit, it would try to reconnect to 'nil'...
r.url = rURL
// Set permissions associated with the route user (if applicable).
// No lock needed since we are already under client lock.
c.setRoutePermissions(opts.Cluster.Permissions)
}
// Check for TLS
if tlsRequired {
// Copy off the config to add in ServerName if we
tlsConfig := opts.Cluster.TLSConfig.Clone()
// If we solicited, we will act like the client, otherwise the server.
if didSolicit {
c.Debugf("Starting TLS route client handshake")
// Specify the ServerName we are expecting.
host, _, _ := net.SplitHostPort(rURL.Host)
tlsConfig.ServerName = host
c.nc = tls.Client(c.nc, tlsConfig)
} else {
c.Debugf("Starting TLS route server handshake")
c.nc = tls.Server(c.nc, tlsConfig)
}
conn := c.nc.(*tls.Conn)
// Setup the timeout
ttl := secondsToDuration(opts.Cluster.TLSTimeout)
time.AfterFunc(ttl, func() { tlsTimeout(c, conn) })
conn.SetReadDeadline(time.Now().Add(ttl))
c.mu.Unlock()
if err := conn.Handshake(); err != nil {
c.Errorf("TLS route handshake error: %v", err)
c.sendErr("Secure Connection - TLS Required")
c.closeConnection(TLSHandshakeError)
return nil
}
// Reset the read deadline
conn.SetReadDeadline(time.Time{})
// Re-Grab lock
c.mu.Lock()
// Verify that the connection did not go away while we released the lock.
if c.nc == nil {
c.mu.Unlock()
return nil
}
}
// Do final client initialization
// Set the Ping timer
c.setPingTimer()
// For routes, the "client" is added to s.routes only when processing
// the INFO protocol, that is much later.
// In the meantime, if the server shutsdown, there would be no reference
// to the client (connection) to be closed, leaving this readLoop
// uinterrupted, causing the Shutdown() to wait indefinitively.
// We need to store the client in a special map, under a special lock.
s.grMu.Lock()
running := s.grRunning
if running {
s.grTmpClients[c.cid] = c
}
s.grMu.Unlock()
if !running {
c.mu.Unlock()
c.setRouteNoReconnectOnClose()
c.closeConnection(ServerShutdown)
return nil
}
// Check for Auth required state for incoming connections.
// Make sure to do this before spinning up readLoop.
if authRequired && !didSolicit {
ttl := secondsToDuration(opts.Cluster.AuthTimeout)
c.setAuthTimer(ttl)
}
// Spin up the read loop.
s.startGoRoutine(func() { c.readLoop() })
// Spin up the write loop.
s.startGoRoutine(c.writeLoop)
if tlsRequired {
c.Debugf("TLS handshake complete")
cs := c.nc.(*tls.Conn).ConnectionState()
c.Debugf("TLS version %s, cipher suite %s", tlsVersion(cs.Version), tlsCipher(cs.CipherSuite))
}
// Queue Connect proto if we solicited the connection.
if didSolicit {
c.Debugf("Route connect msg sent")
c.sendConnect(tlsRequired)
}
// Send our info to the other side.
// Our new version requires dynamic information for accounts and a nonce.
c.sendInfo(infoJSON)
c.mu.Unlock()
c.Noticef("Route connection created")
return c
}
const (
_CRLF_ = "\r\n"
_EMPTY_ = ""
)
func (s *Server) addRoute(c *client, info *Info) (bool, bool) {
id := c.route.remoteID
sendInfo := false
s.mu.Lock()
if !s.running {
s.mu.Unlock()
return false, false
}
remote, exists := s.remotes[id]
if !exists {
s.routes[c.cid] = c
s.remotes[id] = c
c.mu.Lock()
c.route.connectURLs = info.ClientConnectURLs
cid := c.cid
c.mu.Unlock()
// Remove from the temporary map
s.grMu.Lock()
delete(s.grTmpClients, cid)
s.grMu.Unlock()
// we don't need to send if the only route is the one we just accepted.
sendInfo = len(s.routes) > 1
}
s.mu.Unlock()
if exists {
var r *route
c.mu.Lock()
// upgrade to solicited?
if c.route.didSolicit {
// Make a copy
rs := *c.route
r = &rs
}
c.mu.Unlock()
remote.mu.Lock()
// r will be not nil if c.route.didSolicit was true
if r != nil {
// If we upgrade to solicited, we still want to keep the remote's
// connectURLs. So transfer those.
r.connectURLs = remote.route.connectURLs
remote.route = r
}
// This is to mitigate the issue where both sides add the route
// on the opposite connection, and therefore end-up with both
// connections being dropped.
remote.route.retry = true
remote.mu.Unlock()
}
return !exists, sendInfo
}
// updateRouteSubscriptionMap will make sure to update the route map for the subscription. Will
// also forward to all routes if needed.
func (s *Server) updateRouteSubscriptionMap(acc *Account, sub *subscription, delta int32) {
if acc == nil || sub == nil {
return
}
acc.mu.RLock()
rm := acc.rm
acc.mu.RUnlock()
// This is non-nil when we know we are in cluster mode.
if rm == nil {
return
}
// We only store state on local subs for transmission across routes.
if sub.client == nil || sub.client.typ != CLIENT {
return
}
// Create the fast key which will use the subject or 'subject<spc>queue' for queue subscribers.
var (
_rkey [1024]byte
key []byte
qi int
)
if sub.queue != nil {
// Just make the key subject spc group, e.g. 'foo bar'
key = _rkey[:0]
key = append(key, sub.subject...)
key = append(key, byte(' '))
qi = len(key)
key = append(key, sub.queue...)
} else {
key = sub.subject
}
// We always update for a queue subscriber since we need to send our relative weight.
var entry *rme
var ok bool
// Always update if a queue subscriber.
update := qi > 0
// Copy to hold outside acc lock.
var entryN int32
acc.mu.Lock()
if entry, ok = rm[string(key)]; ok {
entry.n += delta
if entry.n <= 0 {
delete(rm, string(key))
update = true // Update for deleting,
}
} else if delta > 0 {
entry = &rme{qi, delta}
rm[string(key)] = entry
update = true // Adding for normal sub means update.
}
if entry != nil {
entryN = entry.n
}
acc.mu.Unlock()
if !update || entry == nil {
return
}
// We need to send out this update.
// If we are sending a queue sub, copy and place in the queue weight.
if sub.queue != nil {
sub.client.mu.Lock()
nsub := *sub
sub.client.mu.Unlock()
nsub.qw = entryN
sub = &nsub
}
// Note that queue unsubs where entry.n > 0 are still
// subscribes with a smaller weight.
if entryN > 0 {
s.broadcastSubscribe(sub)
} else {
s.broadcastUnSubscribe(sub)
}
}
// broadcastSubscribe will forward a client subscription
// to all active routes as needed.
func (s *Server) broadcastSubscribe(sub *subscription) {
trace := atomic.LoadInt32(&s.logging.trace) == 1
s.mu.Lock()
subs := []*subscription{sub}
for _, route := range s.routes {
route.mu.Lock()
route.sendRouteSubProtos(subs, trace, func(sub *subscription) bool {
return route.canImport(string(sub.subject))
})
route.mu.Unlock()
}
s.mu.Unlock()
}
// broadcastUnSubscribe will forward a client unsubscribe
// action to all active routes.
func (s *Server) broadcastUnSubscribe(sub *subscription) {
trace := atomic.LoadInt32(&s.logging.trace) == 1
s.mu.Lock()
subs := []*subscription{sub}
for _, route := range s.routes {
route.mu.Lock()
route.sendRouteUnSubProtos(subs, trace, func(sub *subscription) bool {
return route.canImport(string(sub.subject))
})
route.mu.Unlock()
}
s.mu.Unlock()
}
func (s *Server) routeAcceptLoop(ch chan struct{}) {
defer func() {
if ch != nil {
close(ch)
}
}()
// Snapshot server options.
opts := s.getOpts()
// Snapshot server options.
port := opts.Cluster.Port
if port == -1 {
port = 0
}
hp := net.JoinHostPort(opts.Cluster.Host, strconv.Itoa(port))
l, e := net.Listen("tcp", hp)
if e != nil {
s.Fatalf("Error listening on router port: %d - %v", opts.Cluster.Port, e)
return
}
s.Noticef("Listening for route connections on %s",
net.JoinHostPort(opts.Cluster.Host, strconv.Itoa(l.Addr().(*net.TCPAddr).Port)))
s.mu.Lock()
// For tests, we want to be able to make this server behave
// as an older server so we use the variable which we can override.
proto := testRouteProto
// Check for TLSConfig
tlsReq := opts.Cluster.TLSConfig != nil
info := Info{
ID: s.info.ID,
Version: s.info.Version,
GoVersion: runtime.Version(),
AuthRequired: false,
TLSRequired: tlsReq,
TLSVerify: tlsReq,
MaxPayload: s.info.MaxPayload,
Proto: proto,
}
// Set this if only if advertise is not disabled
if !opts.Cluster.NoAdvertise {
info.ClientConnectURLs = s.clientConnectURLs
}
// If we have selected a random port...
if port == 0 {
// Write resolved port back to options.
opts.Cluster.Port = l.Addr().(*net.TCPAddr).Port
}
// Keep track of actual listen port. This will be needed in case of
// config reload.
s.clusterActualPort = opts.Cluster.Port
// Check for Auth items
if opts.Cluster.Username != "" {
info.AuthRequired = true
}
// Check for permissions.
if opts.Cluster.Permissions != nil {
info.Import = opts.Cluster.Permissions.Import
info.Export = opts.Cluster.Permissions.Export
}
s.routeInfo = info
// Possibly override Host/Port and set IP based on Cluster.Advertise
if err := s.setRouteInfoHostPortAndIP(); err != nil {
s.Fatalf("Error setting route INFO with Cluster.Advertise value of %s, err=%v", s.opts.Cluster.Advertise, err)
l.Close()
s.mu.Unlock()
return
}
// Setup state that can enable shutdown
s.routeListener = l
s.mu.Unlock()
// Let them know we are up
close(ch)
ch = nil
tmpDelay := ACCEPT_MIN_SLEEP
for s.isRunning() {
conn, err := l.Accept()
if err != nil {
if ne, ok := err.(net.Error); ok && ne.Temporary() {
s.Debugf("Temporary Route Accept Errorf(%v), sleeping %dms",
ne, tmpDelay/time.Millisecond)
time.Sleep(tmpDelay)
tmpDelay *= 2
if tmpDelay > ACCEPT_MAX_SLEEP {
tmpDelay = ACCEPT_MAX_SLEEP
}
} else if s.isRunning() {
s.Noticef("Accept error: %v", err)
}
continue
}
tmpDelay = ACCEPT_MIN_SLEEP
s.startGoRoutine(func() {
s.createRoute(conn, nil)
s.grWG.Done()
})
}
s.Debugf("Router accept loop exiting..")
s.done <- true
}
// Similar to setInfoHostPortAndGenerateJSON, but for routeInfo.
func (s *Server) setRouteInfoHostPortAndIP() error {
if s.opts.Cluster.Advertise != "" {
advHost, advPort, err := parseHostPort(s.opts.Cluster.Advertise, s.opts.Cluster.Port)
if err != nil {
return err
}
s.routeInfo.Host = advHost
s.routeInfo.Port = advPort
s.routeInfo.IP = fmt.Sprintf("nats-route://%s/", net.JoinHostPort(advHost, strconv.Itoa(advPort)))
} else {
s.routeInfo.Host = s.opts.Cluster.Host
s.routeInfo.Port = s.opts.Cluster.Port
s.routeInfo.IP = ""
}
// (re)generate the routeInfoJSON byte array
s.generateRouteInfoJSON()
return nil
}
// StartRouting will start the accept loop on the cluster host:port
// and will actively try to connect to listed routes.
func (s *Server) StartRouting(clientListenReady chan struct{}) {
defer s.grWG.Done()
// Wait for the client listen port to be opened, and
// the possible ephemeral port to be selected.
<-clientListenReady
// Spin up the accept loop
ch := make(chan struct{})
go s.routeAcceptLoop(ch)
<-ch
// Solicit Routes if needed.
s.solicitRoutes(s.getOpts().Routes)
}
func (s *Server) reConnectToRoute(rURL *url.URL, rtype RouteType) {
tryForEver := rtype == Explicit
// If A connects to B, and B to A (regardless if explicit or
// implicit - due to auto-discovery), and if each server first
// registers the route on the opposite TCP connection, the
// two connections will end-up being closed.
// Add some random delay to reduce risk of repeated failures.
delay := time.Duration(rand.Intn(100)) * time.Millisecond
if tryForEver {
delay += DEFAULT_ROUTE_RECONNECT
}
select {
case <-time.After(delay):
case <-s.quitCh:
s.grWG.Done()
return
}
s.connectToRoute(rURL, tryForEver)
}
// Checks to make sure the route is still valid.
func (s *Server) routeStillValid(rURL *url.URL) bool {
for _, ri := range s.getOpts().Routes {
if urlsAreEqual(ri, rURL) {
return true
}
}
return false
}
func (s *Server) connectToRoute(rURL *url.URL, tryForEver bool) {
// Snapshot server options.
opts := s.getOpts()
defer s.grWG.Done()
attempts := 0
for s.isRunning() && rURL != nil {
if tryForEver && !s.routeStillValid(rURL) {
return
}
s.Debugf("Trying to connect to route on %s", rURL.Host)
conn, err := net.DialTimeout("tcp", rURL.Host, DEFAULT_ROUTE_DIAL)
if err != nil {
s.Errorf("Error trying to connect to route: %v", err)
if !tryForEver {
if opts.Cluster.ConnectRetries <= 0 {
return
}
attempts++
if attempts > opts.Cluster.ConnectRetries {
return
}
}
select {
case <-s.quitCh:
return
case <-time.After(DEFAULT_ROUTE_CONNECT):
continue
}
}
if tryForEver && !s.routeStillValid(rURL) {
conn.Close()
return
}
// We have a route connection here.
// Go ahead and create it and exit this func.
s.createRoute(conn, rURL)
return
}
}
func (c *client) isSolicitedRoute() bool {
c.mu.Lock()
defer c.mu.Unlock()
return c.typ == ROUTER && c.route != nil && c.route.didSolicit
}
func (s *Server) solicitRoutes(routes []*url.URL) {
for _, r := range routes {
route := r
s.startGoRoutine(func() { s.connectToRoute(route, true) })
}
}
|