/usr/share/doc/ganeti/html/design-2.1.html is in ganeti-doc 2.16.0~rc2-1build1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
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 | <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>Ganeti 2.1 design — Ganeti 2.16.0~rc2 documentation</title>
<link rel="stylesheet" href="_static/style.css" type="text/css" />
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
<script type="text/javascript">
var DOCUMENTATION_OPTIONS = {
URL_ROOT: './',
VERSION: '2.16.0~rc2',
COLLAPSE_INDEX: false,
FILE_SUFFIX: '.html',
HAS_SOURCE: true,
SOURCELINK_SUFFIX: '.txt'
};
</script>
<script type="text/javascript" src="_static/jquery.js"></script>
<script type="text/javascript" src="_static/underscore.js"></script>
<script type="text/javascript" src="_static/doctools.js"></script>
<link rel="search" title="Search" href="search.html" />
<link rel="next" title="Ganeti 2.2 design" href="design-2.2.html" />
<link rel="prev" title="Ganeti 2.0 design" href="design-2.0.html" />
</head>
<body>
<div class="related" role="navigation" aria-label="related navigation">
<h3>Navigation</h3>
<ul>
<li class="right" style="margin-right: 10px">
<a href="design-2.2.html" title="Ganeti 2.2 design"
accesskey="N">next</a></li>
<li class="right" >
<a href="design-2.0.html" title="Ganeti 2.0 design"
accesskey="P">previous</a> |</li>
<li class="nav-item nav-item-0"><a href="index.html">Ganeti 2.16.0~rc2 documentation</a> »</li>
</ul>
</div>
<div class="document">
<div class="documentwrapper">
<div class="bodywrapper">
<div class="body" role="main">
<div class="section" id="ganeti-2-1-design">
<h1><a class="toc-backref" href="#id18">Ganeti 2.1 design</a><a class="headerlink" href="#ganeti-2-1-design" title="Permalink to this headline">¶</a></h1>
<p>This document describes the major changes in Ganeti 2.1 compared to
the 2.0 version.</p>
<p>The 2.1 version will be a relatively small release. Its main aim is to
avoid changing too much of the core code, while addressing issues and
adding new features and improvements over 2.0, in a timely fashion.</p>
<div class="contents topic" id="contents">
<p class="topic-title first">Contents</p>
<ul class="simple">
<li><a class="reference internal" href="#ganeti-2-1-design" id="id18">Ganeti 2.1 design</a><ul>
<li><a class="reference internal" href="#objective" id="id19">Objective</a></li>
<li><a class="reference internal" href="#detailed-design" id="id20">Detailed design</a><ul>
<li><a class="reference internal" href="#core-changes" id="id21">Core changes</a><ul>
<li><a class="reference internal" href="#storage-units-modelling" id="id22">Storage units modelling</a></li>
<li><a class="reference internal" href="#locking-improvements" id="id23">Locking improvements</a></li>
<li><a class="reference internal" href="#node-daemon-availability" id="id24">Node daemon availability</a></li>
</ul>
</li>
<li><a class="reference internal" href="#new-features" id="id25">New Features</a><ul>
<li><a class="reference internal" href="#automated-ganeti-cluster-merger" id="id26">Automated Ganeti Cluster Merger</a></li>
</ul>
</li>
<li><a class="reference internal" href="#feature-changes" id="id27">Feature changes</a><ul>
<li><a class="reference internal" href="#ganeti-confd" id="id28">Ganeti Confd</a></li>
<li><a class="reference internal" href="#redistribute-config" id="id29">Redistribute Config</a></li>
<li><a class="reference internal" href="#vnc-console-password" id="id30">VNC Console Password</a></li>
<li><a class="reference internal" href="#disk-net-parameters" id="id31">Disk/Net parameters</a></li>
<li><a class="reference internal" href="#global-hypervisor-parameters" id="id32">Global hypervisor parameters</a></li>
<li><a class="reference internal" href="#non-bridged-instances-support" id="id33">Non bridged instances support</a></li>
<li><a class="reference internal" href="#introducing-persistent-uuids" id="id34">Introducing persistent UUIDs</a></li>
<li><a class="reference internal" href="#automated-disk-repairs-infrastructure" id="id35">Automated disk repairs infrastructure</a></li>
<li><a class="reference internal" href="#user-id-pool" id="id36">User-id pool</a></li>
</ul>
</li>
<li><a class="reference internal" href="#external-interface-changes" id="id37">External interface changes</a><ul>
<li><a class="reference internal" href="#os-api" id="id38">OS API</a></li>
<li><a class="reference internal" href="#iallocator-changes" id="id39">IAllocator changes</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</div>
<div class="section" id="objective">
<h2><a class="toc-backref" href="#id19">Objective</a><a class="headerlink" href="#objective" title="Permalink to this headline">¶</a></h2>
<p>Ganeti 2.1 will add features to help further automatization of cluster
operations, further improve scalability to even bigger clusters, and
make it easier to debug the Ganeti core.</p>
</div>
<div class="section" id="detailed-design">
<h2><a class="toc-backref" href="#id20">Detailed design</a><a class="headerlink" href="#detailed-design" title="Permalink to this headline">¶</a></h2>
<p>As for 2.0 we divide the 2.1 design into three areas:</p>
<ul class="simple">
<li>core changes, which affect the master daemon/job queue/locking or
all/most logical units</li>
<li>logical unit/feature changes</li>
<li>external interface changes (eg. command line, os api, hooks, …)</li>
</ul>
<div class="section" id="core-changes">
<h3><a class="toc-backref" href="#id21">Core changes</a><a class="headerlink" href="#core-changes" title="Permalink to this headline">¶</a></h3>
<div class="section" id="storage-units-modelling">
<h4><a class="toc-backref" href="#id22">Storage units modelling</a><a class="headerlink" href="#storage-units-modelling" title="Permalink to this headline">¶</a></h4>
<p>Currently, Ganeti has a good model of the block devices for instances
(e.g. LVM logical volumes, files, DRBD devices, etc.) but none of the
storage pools that are providing the space for these front-end
devices. For example, there are hardcoded inter-node RPC calls for
volume group listing, file storage creation/deletion, etc.</p>
<p>The storage units framework will implement a generic handling for all
kinds of storage backends:</p>
<ul class="simple">
<li>LVM physical volumes</li>
<li>LVM volume groups</li>
<li>File-based storage directories</li>
<li>any other future storage method</li>
</ul>
<p>There will be a generic list of methods that each storage unit type
will provide, like:</p>
<ul class="simple">
<li>list of storage units of this type</li>
<li>check status of the storage unit</li>
</ul>
<p>Additionally, there will be specific methods for each method, for
example:</p>
<ul class="simple">
<li>enable/disable allocations on a specific PV</li>
<li>file storage directory creation/deletion</li>
<li>VG consistency fixing</li>
</ul>
<p>This will allow a much better modeling and unification of the various
RPC calls related to backend storage pool in the future. Ganeti 2.1 is
intended to add the basics of the framework, and not necessarilly move
all the curent VG/FileBased operations to it.</p>
<p>Note that while we model both LVM PVs and LVM VGs, the framework will
<strong>not</strong> model any relationship between the different types. In other
words, we model neither inheritances nor stacking, since this is too
complex for our needs. While a <code class="docutils literal"><span class="pre">vgreduce</span></code> operation on a LVM VG
could actually remove a PV from it, this will not be handled at the
framework level, but at individual operation level. The goal is that
this is a lightweight framework, for abstracting the different storage
operation, and not for modelling the storage hierarchy.</p>
</div>
<div class="section" id="locking-improvements">
<h4><a class="toc-backref" href="#id23">Locking improvements</a><a class="headerlink" href="#locking-improvements" title="Permalink to this headline">¶</a></h4>
<div class="section" id="current-state-and-shortcomings">
<h5>Current State and shortcomings<a class="headerlink" href="#current-state-and-shortcomings" title="Permalink to this headline">¶</a></h5>
<p>The class <code class="docutils literal"><span class="pre">LockSet</span></code> (see <code class="docutils literal"><span class="pre">lib/locking.py</span></code>) is a container for one or
many <code class="docutils literal"><span class="pre">SharedLock</span></code> instances. It provides an interface to add/remove
locks and to acquire and subsequently release any number of those locks
contained in it.</p>
<p>Locks in a <code class="docutils literal"><span class="pre">LockSet</span></code> are always acquired in alphabetic order. Due to
the way we’re using locks for nodes and instances (the single cluster
lock isn’t affected by this issue) this can lead to long delays when
acquiring locks if another operation tries to acquire multiple locks but
has to wait for yet another operation.</p>
<p>In the following demonstration we assume to have the instance locks
<code class="docutils literal"><span class="pre">inst1</span></code>, <code class="docutils literal"><span class="pre">inst2</span></code>, <code class="docutils literal"><span class="pre">inst3</span></code> and <code class="docutils literal"><span class="pre">inst4</span></code>.</p>
<ol class="arabic simple">
<li>Operation A grabs lock for instance <code class="docutils literal"><span class="pre">inst4</span></code>.</li>
<li>Operation B wants to acquire all instance locks in alphabetic order,
but it has to wait for <code class="docutils literal"><span class="pre">inst4</span></code>.</li>
<li>Operation C tries to lock <code class="docutils literal"><span class="pre">inst1</span></code>, but it has to wait until
Operation B (which is trying to acquire all locks) releases the lock
again.</li>
<li>Operation A finishes and releases lock on <code class="docutils literal"><span class="pre">inst4</span></code>. Operation B can
continue and eventually releases all locks.</li>
<li>Operation C can get <code class="docutils literal"><span class="pre">inst1</span></code> lock and finishes.</li>
</ol>
<p>Technically there’s no need for Operation C to wait for Operation A, and
subsequently Operation B, to finish. Operation B can’t continue until
Operation A is done (it has to wait for <code class="docutils literal"><span class="pre">inst4</span></code>), anyway.</p>
</div>
<div class="section" id="proposed-changes">
<h5>Proposed changes<a class="headerlink" href="#proposed-changes" title="Permalink to this headline">¶</a></h5>
<div class="section" id="non-blocking-lock-acquiring">
<h6>Non-blocking lock acquiring<a class="headerlink" href="#non-blocking-lock-acquiring" title="Permalink to this headline">¶</a></h6>
<p>Acquiring locks for OpCode execution is always done in blocking mode.
They won’t return until the lock has successfully been acquired (or an
error occurred, although we won’t cover that case here).</p>
<p><code class="docutils literal"><span class="pre">SharedLock</span></code> and <code class="docutils literal"><span class="pre">LockSet</span></code> must be able to be acquired in a
non-blocking way. They must support a timeout and abort trying to
acquire the lock(s) after the specified amount of time.</p>
</div>
<div class="section" id="retry-acquiring-locks">
<h6>Retry acquiring locks<a class="headerlink" href="#retry-acquiring-locks" title="Permalink to this headline">¶</a></h6>
<p>To prevent other operations from waiting for a long time, such as
described in the demonstration before, <code class="docutils literal"><span class="pre">LockSet</span></code> must not keep locks
for a prolonged period of time when trying to acquire two or more locks.
Instead it should, with an increasing timeout for acquiring all locks,
release all locks again and sleep some time if it fails to acquire all
requested locks.</p>
<p>A good timeout value needs to be determined. In any case should
<code class="docutils literal"><span class="pre">LockSet</span></code> proceed to acquire locks in blocking mode after a few
(unsuccessful) attempts to acquire all requested locks.</p>
<p>One proposal for the timeout is to use <code class="docutils literal"><span class="pre">2**tries</span></code> seconds, where
<code class="docutils literal"><span class="pre">tries</span></code> is the number of unsuccessful tries.</p>
<p>In the demonstration before this would allow Operation C to continue
after Operation B unsuccessfully tried to acquire all locks and released
all acquired locks (<code class="docutils literal"><span class="pre">inst1</span></code>, <code class="docutils literal"><span class="pre">inst2</span></code> and <code class="docutils literal"><span class="pre">inst3</span></code>) again.</p>
</div>
</div>
<div class="section" id="other-solutions-discussed">
<h5>Other solutions discussed<a class="headerlink" href="#other-solutions-discussed" title="Permalink to this headline">¶</a></h5>
<p>There was also some discussion on going one step further and extend the
job queue (see <code class="docutils literal"><span class="pre">lib/jqueue.py</span></code>) to select the next task for a worker
depending on whether it can acquire the necessary locks. While this may
reduce the number of necessary worker threads and/or increase throughput
on large clusters with many jobs, it also brings many potential
problems, such as contention and increased memory usage, with it. As
this would be an extension of the changes proposed before it could be
implemented at a later point in time, but we decided to stay with the
simpler solution for now.</p>
</div>
<div class="section" id="implementation-details">
<h5>Implementation details<a class="headerlink" href="#implementation-details" title="Permalink to this headline">¶</a></h5>
<div class="section" id="sharedlock-redesign">
<h6><code class="docutils literal"><span class="pre">SharedLock</span></code> redesign<a class="headerlink" href="#sharedlock-redesign" title="Permalink to this headline">¶</a></h6>
<p>The current design of <code class="docutils literal"><span class="pre">SharedLock</span></code> is not good for supporting timeouts
when acquiring a lock and there are also minor fairness issues in it. We
plan to address both with a redesign. A proof of concept implementation
was written and resulted in significantly simpler code.</p>
<p>Currently <code class="docutils literal"><span class="pre">SharedLock</span></code> uses two separate queues for shared and
exclusive acquires and waiters get to run in turns. This means if an
exclusive acquire is released, the lock will allow shared waiters to run
and vice versa. Although it’s still fair in the end there is a slight
bias towards shared waiters in the current implementation. The same
implementation with two shared queues can not support timeouts without
adding a lot of complexity.</p>
<p>Our proposed redesign changes <code class="docutils literal"><span class="pre">SharedLock</span></code> to have only one single
queue. There will be one condition (see <a class="reference internal" href="#condition">Condition</a> for a note about
performance) in the queue per exclusive acquire and two for all shared
acquires (see below for an explanation). The maximum queue length will
always be <code class="docutils literal"><span class="pre">2</span> <span class="pre">+</span> <span class="pre">(number</span> <span class="pre">of</span> <span class="pre">exclusive</span> <span class="pre">acquires</span> <span class="pre">waiting)</span></code>. The number of
queue entries for shared acquires can vary from 0 to 2.</p>
<p>The two conditions for shared acquires are a bit special. They will be
used in turn. When the lock is instantiated, no conditions are in the
queue. As soon as the first shared acquire arrives (and there are
holder(s) or waiting acquires; see <a class="reference internal" href="#acquire">Acquire</a>), the active condition is
added to the queue. Until it becomes the topmost condition in the queue
and has been notified, any shared acquire is added to this active
condition. When the active condition is notified, the conditions are
swapped and further shared acquires are added to the previously inactive
condition (which has now become the active condition). After all waiters
on the previously active (now inactive) and now notified condition
received the notification, it is removed from the queue of pending
acquires.</p>
<p>This means shared acquires will skip any exclusive acquire in the queue.
We believe it’s better to improve parallelization on operations only
asking for shared (or read-only) locks. Exclusive operations holding the
same lock can not be parallelized.</p>
<div class="section" id="acquire">
<h7>Acquire<a class="headerlink" href="#acquire" title="Permalink to this headline">¶</a></h7>
<p>For exclusive acquires a new condition is created and appended to the
queue. Shared acquires are added to the active condition for shared
acquires and if the condition is not yet on the queue, it’s appended.</p>
<p>The next step is to wait for our condition to be on the top of the queue
(to guarantee fairness). If the timeout expired, we return to the caller
without acquiring the lock. On every notification we check whether the
lock has been deleted, in which case an error is returned to the caller.</p>
<p>The lock can be acquired if we’re on top of the queue (there is no one
else ahead of us). For an exclusive acquire, there must not be other
exclusive or shared holders. For a shared acquire, there must not be an
exclusive holder. If these conditions are all true, the lock is
acquired and we return to the caller. In any other case we wait again on
the condition.</p>
<p>If it was the last waiter on a condition, the condition is removed from
the queue.</p>
<p>Optimization: There’s no need to touch the queue if there are no pending
acquires and no current holders. The caller can have the lock
immediately.</p>
<img src="_images/graphviz-64b42726970b28107274fabd4d6540b8c141739b.png" alt="digraph "design-2.1-lock-acquire" {
graph[fontsize=8, fontname="Helvetica"]
node[fontsize=8, fontname="Helvetica", width="0", height="0"]
edge[fontsize=8, fontname="Helvetica"]
/* Actions */
abort[label="Abort\n(couldn't acquire)"]
acquire[label="Acquire lock"]
add_to_queue[label="Add condition to queue"]
wait[label="Wait for notification"]
remove_from_queue[label="Remove from queue"]
/* Conditions */
alone[label="Empty queue\nand can acquire?", shape=diamond]
have_timeout[label="Do I have\ntimeout?", shape=diamond]
top_of_queue_and_can_acquire[
label="On top of queue and\ncan acquire lock?",
shape=diamond,
]
/* Lines */
alone->acquire[label="Yes"]
alone->add_to_queue[label="No"]
have_timeout->abort[label="Yes"]
have_timeout->wait[label="No"]
top_of_queue_and_can_acquire->acquire[label="Yes"]
top_of_queue_and_can_acquire->have_timeout[label="No"]
add_to_queue->wait
wait->top_of_queue_and_can_acquire
acquire->remove_from_queue
}" />
</div>
<div class="section" id="release">
<h7>Release<a class="headerlink" href="#release" title="Permalink to this headline">¶</a></h7>
<p>First the lock removes the caller from the internal owner list. If there
are pending acquires in the queue, the first (the oldest) condition is
notified.</p>
<p>If the first condition was the active condition for shared acquires, the
inactive condition will be made active. This ensures fairness with
exclusive locks by forcing consecutive shared acquires to wait in the
queue.</p>
<img src="_images/graphviz-2f245c2527e06b4204f59547cdc067985aab10cd.png" alt="digraph "design-2.1-lock-release" {
graph[fontsize=8, fontname="Helvetica"]
node[fontsize=8, fontname="Helvetica", width="0", height="0"]
edge[fontsize=8, fontname="Helvetica"]
/* Actions */
remove_from_owners[label="Remove from owner list"]
notify[label="Notify topmost"]
swap_shared[label="Swap shared conditions"]
success[label="Success"]
/* Conditions */
have_pending[label="Any pending\nacquires?", shape=diamond]
was_active_queue[
label="Was active condition\nfor shared acquires?",
shape=diamond,
]
/* Lines */
remove_from_owners->have_pending
have_pending->notify[label="Yes"]
have_pending->success[label="No"]
notify->was_active_queue
was_active_queue->swap_shared[label="Yes"]
was_active_queue->success[label="No"]
swap_shared->success
}" />
</div>
<div class="section" id="delete">
<h7>Delete<a class="headerlink" href="#delete" title="Permalink to this headline">¶</a></h7>
<p>The caller must either hold the lock in exclusive mode already or the
lock must be acquired in exclusive mode. Trying to delete a lock while
it’s held in shared mode must fail.</p>
<p>After ensuring the lock is held in exclusive mode, the lock will mark
itself as deleted and continue to notify all pending acquires. They will
wake up, notice the deleted lock and return an error to the caller.</p>
</div>
</div>
<div class="section" id="condition">
<h6>Condition<a class="headerlink" href="#condition" title="Permalink to this headline">¶</a></h6>
<p>Note: This is not necessary for the locking changes above, but it may be
a good optimization (pending performance tests).</p>
<p>The existing locking code in Ganeti 2.0 uses Python’s built-in
<code class="docutils literal"><span class="pre">threading.Condition</span></code> class. Unfortunately <code class="docutils literal"><span class="pre">Condition</span></code> implements
timeouts by sleeping 1ms to 20ms between tries to acquire the condition
lock in non-blocking mode. This requires unnecessary context switches
and contention on the CPython GIL (Global Interpreter Lock).</p>
<p>By using POSIX pipes (see <code class="docutils literal"><span class="pre">pipe(2)</span></code>) we can use the operating system’s
support for timeouts on file descriptors (see <code class="docutils literal"><span class="pre">select(2)</span></code>). A custom
condition class will have to be written for this.</p>
<p>On instantiation the class creates a pipe. After each notification the
previous pipe is abandoned and re-created (technically the old pipe
needs to stay around until all notifications have been delivered).</p>
<p>All waiting clients of the condition use <code class="docutils literal"><span class="pre">select(2)</span></code> or <code class="docutils literal"><span class="pre">poll(2)</span></code> to
wait for notifications, optionally with a timeout. A notification will
be signalled to the waiting clients by closing the pipe. If the pipe
wasn’t closed during the timeout, the waiting function returns to its
caller nonetheless.</p>
</div>
</div>
</div>
<div class="section" id="node-daemon-availability">
<h4><a class="toc-backref" href="#id24">Node daemon availability</a><a class="headerlink" href="#node-daemon-availability" title="Permalink to this headline">¶</a></h4>
<div class="section" id="id1">
<h5>Current State and shortcomings<a class="headerlink" href="#id1" title="Permalink to this headline">¶</a></h5>
<p>Currently, when a Ganeti node suffers serious system disk damage, the
migration/failover of an instance may not correctly shutdown the virtual
machine on the broken node causing instances duplication. The <code class="docutils literal"><span class="pre">gnt-node</span>
<span class="pre">powercycle</span></code> command can be used to force a node reboot and thus to
avoid duplicated instances. This command relies on node daemon
availability, though, and thus can fail if the node daemon has some
pages swapped out of ram, for example.</p>
</div>
<div class="section" id="id2">
<h5>Proposed changes<a class="headerlink" href="#id2" title="Permalink to this headline">¶</a></h5>
<p>The proposed solution forces node daemon to run exclusively in RAM. It
uses python ctypes to to call <code class="docutils literal"><span class="pre">mlockall(MCL_CURRENT</span> <span class="pre">|</span> <span class="pre">MCL_FUTURE)</span></code> on
the node daemon process and all its children. In addition another log
handler has been implemented for node daemon to redirect to
<code class="docutils literal"><span class="pre">/dev/console</span></code> messages that cannot be written on the logfile.</p>
<p>With these changes node daemon can successfully run basic tasks such as
a powercycle request even when the system disk is heavily damaged and
reading/writing to disk fails constantly.</p>
</div>
</div>
</div>
<div class="section" id="new-features">
<h3><a class="toc-backref" href="#id25">New Features</a><a class="headerlink" href="#new-features" title="Permalink to this headline">¶</a></h3>
<div class="section" id="automated-ganeti-cluster-merger">
<h4><a class="toc-backref" href="#id26">Automated Ganeti Cluster Merger</a><a class="headerlink" href="#automated-ganeti-cluster-merger" title="Permalink to this headline">¶</a></h4>
<div class="section" id="current-situation">
<h5>Current situation<a class="headerlink" href="#current-situation" title="Permalink to this headline">¶</a></h5>
<p>Currently there’s no easy way to merge two or more clusters together.
But in order to optimize resources this is a needed missing piece. The
goal of this design doc is to come up with a easy to use solution which
allows you to merge two or more clusters together.</p>
</div>
<div class="section" id="initial-contact">
<h5>Initial contact<a class="headerlink" href="#initial-contact" title="Permalink to this headline">¶</a></h5>
<p>As the design of Ganeti is based on an autonomous system, Ganeti by
itself has no way to reach nodes outside of its cluster. To overcome
this situation we’re required to prepare the cluster before we can go
ahead with the actual merge: We’ve to replace at least the ssh keys on
the affected nodes before we can do any operation within <code class="docutils literal"><span class="pre">gnt-</span></code>
commands.</p>
<p>To make this a automated process we’ll ask the user to provide us with
the root password of every cluster we’ve to merge. We use the password
to grab the current <code class="docutils literal"><span class="pre">id_dsa</span></code> key and then rely on that ssh key for any
further communication to be made until the cluster is fully merged.</p>
</div>
<div class="section" id="cluster-merge">
<h5>Cluster merge<a class="headerlink" href="#cluster-merge" title="Permalink to this headline">¶</a></h5>
<p>After initial contact we do the cluster merge:</p>
<ol class="arabic simple">
<li>Grab the list of nodes</li>
<li>On all nodes add our own <code class="docutils literal"><span class="pre">id_dsa.pub</span></code> key to <code class="docutils literal"><span class="pre">authorized_keys</span></code></li>
<li>Stop all instances running on the merging cluster</li>
<li>Disable <code class="docutils literal"><span class="pre">ganeti-watcher</span></code> as it tries to restart Ganeti daemons</li>
<li>Stop all Ganeti daemons on all merging nodes</li>
<li>Grab the <code class="docutils literal"><span class="pre">config.data</span></code> from the master of the merging cluster</li>
<li>Stop local <code class="docutils literal"><span class="pre">ganeti-masterd</span></code></li>
<li>Merge the config:<ol class="arabic">
<li>Open our own cluster <code class="docutils literal"><span class="pre">config.data</span></code></li>
<li>Open cluster <code class="docutils literal"><span class="pre">config.data</span></code> of the merging cluster</li>
<li>Grab all nodes of the merging cluster</li>
<li>Set <code class="docutils literal"><span class="pre">master_candidate</span></code> to false on all merging nodes</li>
<li>Add the nodes to our own cluster <code class="docutils literal"><span class="pre">config.data</span></code></li>
<li>Grab all the instances on the merging cluster</li>
<li>Adjust the port if the instance has drbd layout:<ol class="arabic">
<li>In <code class="docutils literal"><span class="pre">logical_id</span></code> (index 2)</li>
<li>In <code class="docutils literal"><span class="pre">physical_id</span></code> (index 1 and 3)</li>
</ol>
</li>
<li>Add the instances to our own cluster <code class="docutils literal"><span class="pre">config.data</span></code></li>
</ol>
</li>
<li>Start <code class="docutils literal"><span class="pre">ganeti-masterd</span></code> with <code class="docutils literal"><span class="pre">--no-voting</span></code> <code class="docutils literal"><span class="pre">--yes-do-it</span></code></li>
<li><code class="docutils literal"><span class="pre">gnt-node</span> <span class="pre">add</span> <span class="pre">--readd</span></code> on all merging nodes</li>
<li><code class="docutils literal"><span class="pre">gnt-cluster</span> <span class="pre">redist-conf</span></code></li>
<li>Restart <code class="docutils literal"><span class="pre">ganeti-masterd</span></code> normally</li>
<li>Enable <code class="docutils literal"><span class="pre">ganeti-watcher</span></code> again</li>
<li>Start all merging instances again</li>
</ol>
</div>
<div class="section" id="rollback">
<h5>Rollback<a class="headerlink" href="#rollback" title="Permalink to this headline">¶</a></h5>
<p>Until we actually (re)add any nodes we can abort and rollback the merge
at any point. After merging the config, though, we’ve to get the backup
copy of <code class="docutils literal"><span class="pre">config.data</span></code> (from another master candidate node). And for
security reasons it’s a good idea to undo <code class="docutils literal"><span class="pre">id_dsa.pub</span></code> distribution by
going on every affected node and remove the <code class="docutils literal"><span class="pre">id_dsa.pub</span></code> key again.
Also we’ve to keep in mind, that we’ve to start the Ganeti daemons and
starting up the instances again.</p>
</div>
<div class="section" id="verification">
<h5>Verification<a class="headerlink" href="#verification" title="Permalink to this headline">¶</a></h5>
<p>Last but not least we should verify that the merge was successful.
Therefore we run <code class="docutils literal"><span class="pre">gnt-cluster</span> <span class="pre">verify</span></code>, which ensures that the cluster
overall is in a healthy state. Additional it’s also possible to compare
the list of instances/nodes with a list made prior to the upgrade to
make sure we didn’t lose any data/instance/node.</p>
</div>
<div class="section" id="appendix">
<h5>Appendix<a class="headerlink" href="#appendix" title="Permalink to this headline">¶</a></h5>
<div class="section" id="cluster-merge-py">
<h6>cluster-merge.py<a class="headerlink" href="#cluster-merge-py" title="Permalink to this headline">¶</a></h6>
<p>Used to merge the cluster config. This is a POC and might differ from
actual production code.</p>
<div class="highlight-default"><div class="highlight"><pre><span></span><span class="ch">#!/usr/bin/python</span>
<span class="kn">import</span> <span class="nn">sys</span>
<span class="kn">from</span> <span class="nn">ganeti</span> <span class="k">import</span> <span class="n">config</span>
<span class="kn">from</span> <span class="nn">ganeti</span> <span class="k">import</span> <span class="n">constants</span>
<span class="n">c_mine</span> <span class="o">=</span> <span class="n">config</span><span class="o">.</span><span class="n">ConfigWriter</span><span class="p">(</span><span class="n">offline</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="n">c_other</span> <span class="o">=</span> <span class="n">config</span><span class="o">.</span><span class="n">ConfigWriter</span><span class="p">(</span><span class="n">sys</span><span class="o">.</span><span class="n">argv</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span>
<span class="n">fake_id</span> <span class="o">=</span> <span class="mi">0</span>
<span class="k">for</span> <span class="n">node</span> <span class="ow">in</span> <span class="n">c_other</span><span class="o">.</span><span class="n">GetNodeList</span><span class="p">():</span>
<span class="n">node_info</span> <span class="o">=</span> <span class="n">c_other</span><span class="o">.</span><span class="n">GetNodeInfo</span><span class="p">(</span><span class="n">node</span><span class="p">)</span>
<span class="n">node_info</span><span class="o">.</span><span class="n">master_candidate</span> <span class="o">=</span> <span class="kc">False</span>
<span class="n">c_mine</span><span class="o">.</span><span class="n">AddNode</span><span class="p">(</span><span class="n">node_info</span><span class="p">,</span> <span class="nb">str</span><span class="p">(</span><span class="n">fake_id</span><span class="p">))</span>
<span class="n">fake_id</span> <span class="o">+=</span> <span class="mi">1</span>
<span class="k">for</span> <span class="n">instance</span> <span class="ow">in</span> <span class="n">c_other</span><span class="o">.</span><span class="n">GetInstanceList</span><span class="p">():</span>
<span class="n">instance_info</span> <span class="o">=</span> <span class="n">c_other</span><span class="o">.</span><span class="n">GetInstanceInfo</span><span class="p">(</span><span class="n">instance</span><span class="p">)</span>
<span class="k">for</span> <span class="n">dsk</span> <span class="ow">in</span> <span class="n">instance_info</span><span class="o">.</span><span class="n">disks</span><span class="p">:</span>
<span class="k">if</span> <span class="n">dsk</span><span class="o">.</span><span class="n">dev_type</span> <span class="ow">in</span> <span class="n">constants</span><span class="o">.</span><span class="n">LDS_DRBD</span><span class="p">:</span>
<span class="n">port</span> <span class="o">=</span> <span class="n">c_mine</span><span class="o">.</span><span class="n">AllocatePort</span><span class="p">()</span>
<span class="n">logical_id</span> <span class="o">=</span> <span class="nb">list</span><span class="p">(</span><span class="n">dsk</span><span class="o">.</span><span class="n">logical_id</span><span class="p">)</span>
<span class="n">logical_id</span><span class="p">[</span><span class="mi">2</span><span class="p">]</span> <span class="o">=</span> <span class="n">port</span>
<span class="n">dsk</span><span class="o">.</span><span class="n">logical_id</span> <span class="o">=</span> <span class="nb">tuple</span><span class="p">(</span><span class="n">logical_id</span><span class="p">)</span>
<span class="n">physical_id</span> <span class="o">=</span> <span class="nb">list</span><span class="p">(</span><span class="n">dsk</span><span class="o">.</span><span class="n">physical_id</span><span class="p">)</span>
<span class="n">physical_id</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span> <span class="o">=</span> <span class="n">physical_id</span><span class="p">[</span><span class="mi">3</span><span class="p">]</span> <span class="o">=</span> <span class="n">port</span>
<span class="n">dsk</span><span class="o">.</span><span class="n">physical_id</span> <span class="o">=</span> <span class="nb">tuple</span><span class="p">(</span><span class="n">physical_id</span><span class="p">)</span>
<span class="n">c_mine</span><span class="o">.</span><span class="n">AddInstance</span><span class="p">(</span><span class="n">instance_info</span><span class="p">,</span> <span class="nb">str</span><span class="p">(</span><span class="n">fake_id</span><span class="p">))</span>
<span class="n">fake_id</span> <span class="o">+=</span> <span class="mi">1</span>
</pre></div>
</div>
</div>
</div>
</div>
</div>
<div class="section" id="feature-changes">
<h3><a class="toc-backref" href="#id27">Feature changes</a><a class="headerlink" href="#feature-changes" title="Permalink to this headline">¶</a></h3>
<div class="section" id="ganeti-confd">
<h4><a class="toc-backref" href="#id28">Ganeti Confd</a><a class="headerlink" href="#ganeti-confd" title="Permalink to this headline">¶</a></h4>
<div class="section" id="id3">
<h5>Current State and shortcomings<a class="headerlink" href="#id3" title="Permalink to this headline">¶</a></h5>
<p>In Ganeti 2.0 all nodes are equal, but some are more equal than others.
In particular they are divided between “master”, “master candidates” and
“normal”. (Moreover they can be offline or drained, but this is not
important for the current discussion). In general the whole
configuration is only replicated to master candidates, and some partial
information is spread to all nodes via ssconf.</p>
<p>This change was done so that the most frequent Ganeti operations didn’t
need to contact all nodes, and so clusters could become bigger. If we
want more information to be available on all nodes, we need to add more
ssconf values, which is counter-balancing the change, or to talk with
the master node, which is not designed to happen now, and requires its
availability.</p>
<p>Information such as the instance->primary_node mapping will be needed on
all nodes, and we also want to make sure services external to the
cluster can query this information as well. This information must be
available at all times, so we can’t query it through RAPI, which would
be a single point of failure, as it’s only available on the master.</p>
</div>
<div class="section" id="id4">
<h5>Proposed changes<a class="headerlink" href="#id4" title="Permalink to this headline">¶</a></h5>
<p>In order to allow fast and highly available access read-only to some
configuration values, we’ll create a new ganeti-confd daemon, which will
run on master candidates. This daemon will talk via UDP, and
authenticate messages using HMAC with a cluster-wide shared key. This
key will be generated at cluster init time, and stored on the clusters
alongside the ganeti SSL keys, and readable only by root.</p>
<p>An interested client can query a value by making a request to a subset
of the cluster master candidates. It will then wait to get a few
responses, and use the one with the highest configuration serial number.
Since the configuration serial number is increased each time the ganeti
config is updated, and the serial number is included in all answers,
this can be used to make sure to use the most recent answer, in case
some master candidates are stale or in the middle of a configuration
update.</p>
<p>In order to prevent replay attacks queries will contain the current unix
timestamp according to the client, and the server will verify that its
timestamp is in the same 5 minutes range (this requires synchronized
clocks, which is a good idea anyway). Queries will also contain a “salt”
which they expect the answers to be sent with, and clients are supposed
to accept only answers which contain salt generated by them.</p>
<p>The configuration daemon will be able to answer simple queries such as:</p>
<ul class="simple">
<li>master candidates list</li>
<li>master node</li>
<li>offline nodes</li>
<li>instance list</li>
<li>instance primary nodes</li>
</ul>
<div class="section" id="wire-protocol">
<h6>Wire protocol<a class="headerlink" href="#wire-protocol" title="Permalink to this headline">¶</a></h6>
<p>A confd query will look like this, on the wire:</p>
<div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">plj0</span><span class="p">{</span>
<span class="s2">"msg"</span><span class="p">:</span> <span class="s2">"{</span><span class="se">\"</span><span class="s2">type</span><span class="se">\"</span><span class="s2">: 1,</span>
\<span class="s2">"rsalt</span><span class="se">\"</span><span class="s2">: </span><span class="se">\"</span><span class="s2">9aa6ce92-8336-11de-af38-001d093e835f</span><span class="se">\"</span><span class="s2">,</span>
\<span class="s2">"protocol</span><span class="se">\"</span><span class="s2">: 1,</span>
\<span class="s2">"query</span><span class="se">\"</span><span class="s2">: </span><span class="se">\"</span><span class="s2">node1.example.com</span><span class="se">\"</span><span class="s2">}</span><span class="se">\n</span><span class="s2">"</span><span class="p">,</span>
<span class="s2">"salt"</span><span class="p">:</span> <span class="s2">"1249637704"</span><span class="p">,</span>
<span class="s2">"hmac"</span><span class="p">:</span> <span class="s2">"4a4139b2c3c5921f7e439469a0a45ad200aead0f"</span>
<span class="p">}</span>
</pre></div>
</div>
<p><code class="docutils literal"><span class="pre">plj0</span></code> is a fourcc that details the message content. It stands for plain
json 0, and can be changed as we move on to different type of protocols
(for example protocol buffers, or encrypted json). What follows is a
json encoded string, with the following fields:</p>
<ul class="simple">
<li><code class="docutils literal"><span class="pre">msg</span></code> contains a JSON-encoded query, its fields are:<ul>
<li><code class="docutils literal"><span class="pre">protocol</span></code>, integer, is the confd protocol version (initially
just <code class="docutils literal"><span class="pre">constants.CONFD_PROTOCOL_VERSION</span></code>, with a value of 1)</li>
<li><code class="docutils literal"><span class="pre">type</span></code>, integer, is the query type. For example “node role by
name” or “node primary ip by instance ip”. Constants will be
provided for the actual available query types</li>
<li><code class="docutils literal"><span class="pre">query</span></code> is a multi-type field (depending on the <code class="docutils literal"><span class="pre">type</span></code> field):<ul>
<li>it can be missing, when the request is fully determined by the
<code class="docutils literal"><span class="pre">type</span></code> field</li>
<li>it can contain a string which denotes the search key: for
example an IP, or a node name</li>
<li>it can contain a dictionary, in which case the actual details
vary further per request type</li>
</ul>
</li>
<li><code class="docutils literal"><span class="pre">rsalt</span></code>, string, is the required response salt; the client must
use it to recognize which answer it’s getting.</li>
</ul>
</li>
<li><code class="docutils literal"><span class="pre">salt</span></code> must be the current unix timestamp, according to the
client; servers should refuse messages which have a wrong timing,
according to their configuration and clock</li>
<li><code class="docutils literal"><span class="pre">hmac</span></code> is an hmac signature of salt+msg, with the cluster hmac key</li>
</ul>
<p>If an answer comes back (which is optional, since confd works over UDP)
it will be in this format:</p>
<div class="highlight-default"><div class="highlight"><pre><span></span><span class="n">plj0</span><span class="p">{</span>
<span class="s2">"msg"</span><span class="p">:</span> <span class="s2">"{</span><span class="se">\"</span><span class="s2">status</span><span class="se">\"</span><span class="s2">: 0,</span>
\<span class="s2">"answer</span><span class="se">\"</span><span class="s2">: 0,</span>
\<span class="s2">"serial</span><span class="se">\"</span><span class="s2">: 42,</span>
\<span class="s2">"protocol</span><span class="se">\"</span><span class="s2">: 1}</span><span class="se">\n</span><span class="s2">"</span><span class="p">,</span>
<span class="s2">"salt"</span><span class="p">:</span> <span class="s2">"9aa6ce92-8336-11de-af38-001d093e835f"</span><span class="p">,</span>
<span class="s2">"hmac"</span><span class="p">:</span> <span class="s2">"aaeccc0dff9328fdf7967cb600b6a80a6a9332af"</span>
<span class="p">}</span>
</pre></div>
</div>
<p>Where:</p>
<ul class="simple">
<li><code class="docutils literal"><span class="pre">plj0</span></code> the message type magic fourcc, as discussed above</li>
<li><code class="docutils literal"><span class="pre">msg</span></code> contains a JSON-encoded answer, its fields are:<ul>
<li><code class="docutils literal"><span class="pre">protocol</span></code>, integer, is the confd protocol version (initially
just constants.CONFD_PROTOCOL_VERSION, with a value of 1)</li>
<li><code class="docutils literal"><span class="pre">status</span></code>, integer, is the error code; initially just <code class="docutils literal"><span class="pre">0</span></code> for
‘ok’ or <code class="docutils literal"><span class="pre">1</span></code> for ‘error’ (in which case answer contains an error
detail, rather than an answer), but in the future it may be
expanded to have more meanings (e.g. <code class="docutils literal"><span class="pre">2</span></code> if the answer is
compressed)</li>
<li><code class="docutils literal"><span class="pre">answer</span></code>, is the actual answer; its type and meaning is query
specific: for example for “node primary ip by instance ip” queries
it will be a string containing an IP address, for “node role by
name” queries it will be an integer which encodes the role
(master, candidate, drained, offline) according to constants</li>
</ul>
</li>
<li><code class="docutils literal"><span class="pre">salt</span></code> is the requested salt from the query; a client can use it
to recognize what query the answer is answering.</li>
<li><code class="docutils literal"><span class="pre">hmac</span></code> is an hmac signature of salt+msg, with the cluster hmac key</li>
</ul>
</div>
</div>
</div>
<div class="section" id="redistribute-config">
<h4><a class="toc-backref" href="#id29">Redistribute Config</a><a class="headerlink" href="#redistribute-config" title="Permalink to this headline">¶</a></h4>
<div class="section" id="id5">
<h5>Current State and shortcomings<a class="headerlink" href="#id5" title="Permalink to this headline">¶</a></h5>
<p>Currently LUClusterRedistConf triggers a copy of the updated
configuration file to all master candidates and of the ssconf files to
all nodes. There are other files which are maintained manually but which
are important to keep in sync. These are:</p>
<ul class="simple">
<li>rapi SSL key certificate file (rapi.pem) (on master candidates)</li>
<li>rapi user/password file rapi_users (on master candidates)</li>
</ul>
<p>Furthermore there are some files which are hypervisor specific but we
may want to keep in sync:</p>
<ul class="simple">
<li>the xen-hvm hypervisor uses one shared file for all vnc passwords, and
copies the file once, during node add. This design is subject to
revision to be able to have different passwords for different groups
of instances via the use of hypervisor parameters, and to allow
xen-hvm and kvm to use an equal system to provide password-protected
vnc sessions. In general, though, it would be useful if the vnc
password files were copied as well, to avoid unwanted vnc password
changes on instance failover/migrate.</li>
</ul>
<p>Optionally the admin may want to also ship files such as the global
xend.conf file, and the network scripts to all nodes.</p>
</div>
<div class="section" id="id6">
<h5>Proposed changes<a class="headerlink" href="#id6" title="Permalink to this headline">¶</a></h5>
<p>RedistributeConfig will be changed to copy also the rapi files, and to
call every enabled hypervisor asking for a list of additional files to
copy. Users will have the possibility to populate a file containing a
list of files to be distributed; this file will be propagated as well.
Such solution is really simple to implement and it’s easily usable by
scripts.</p>
<p>This code will be also shared (via tasklets or by other means, if
tasklets are not ready for 2.1) with the AddNode and SetNodeParams LUs
(so that the relevant files will be automatically shipped to new master
candidates as they are set).</p>
</div>
</div>
<div class="section" id="vnc-console-password">
<h4><a class="toc-backref" href="#id30">VNC Console Password</a><a class="headerlink" href="#vnc-console-password" title="Permalink to this headline">¶</a></h4>
<div class="section" id="id7">
<h5>Current State and shortcomings<a class="headerlink" href="#id7" title="Permalink to this headline">¶</a></h5>
<p>Currently just the xen-hvm hypervisor supports setting a password to
connect the the instances’ VNC console, and has one common password
stored in a file.</p>
<p>This doesn’t allow different passwords for different instances/groups of
instances, and makes it necessary to remember to copy the file around
the cluster when the password changes.</p>
</div>
<div class="section" id="id8">
<h5>Proposed changes<a class="headerlink" href="#id8" title="Permalink to this headline">¶</a></h5>
<p>We’ll change the VNC password file to a vnc_password_file hypervisor
parameter. This way it can have a cluster default, but also a different
value for each instance. The VNC enabled hypervisors (xen and kvm) will
publish all the password files in use through the cluster so that a
redistribute-config will ship them to all nodes (see the Redistribute
Config proposed changes above).</p>
<p>The current VNC_PASSWORD_FILE constant will be removed, but its value
will be used as the default HV_VNC_PASSWORD_FILE value, thus retaining
backwards compatibility with 2.0.</p>
<p>The code to export the list of VNC password files from the hypervisors
to RedistributeConfig will be shared between the KVM and xen-hvm
hypervisors.</p>
</div>
</div>
<div class="section" id="disk-net-parameters">
<h4><a class="toc-backref" href="#id31">Disk/Net parameters</a><a class="headerlink" href="#disk-net-parameters" title="Permalink to this headline">¶</a></h4>
<div class="section" id="id9">
<h5>Current State and shortcomings<a class="headerlink" href="#id9" title="Permalink to this headline">¶</a></h5>
<p>Currently disks and network interfaces have a few tweakable options and
all the rest is left to a default we chose. We’re finding that we need
more and more to tweak some of these parameters, for example to disable
barriers for DRBD devices, or allow striping for the LVM volumes.</p>
<p>Moreover for many of these parameters it will be nice to have
cluster-wide defaults, and then be able to change them per
disk/interface.</p>
</div>
<div class="section" id="id10">
<h5>Proposed changes<a class="headerlink" href="#id10" title="Permalink to this headline">¶</a></h5>
<p>We will add new cluster level diskparams and netparams, which will
contain all the tweakable parameters. All values which have a sensible
cluster-wide default will go into this new structure while parameters
which have unique values will not.</p>
<dl class="docutils">
<dt>Example of network parameters:</dt>
<dd><ul class="first last simple">
<li>mode: bridge/route</li>
<li>link: for mode “bridge” the bridge to connect to, for mode route it
can contain the routing table, or the destination interface</li>
</ul>
</dd>
<dt>Example of disk parameters:</dt>
<dd><ul class="first last simple">
<li>stripe: lvm stripes</li>
<li>stripe_size: lvm stripe size</li>
<li>meta_flushes: drbd, enable/disable metadata “barriers”</li>
<li>data_flushes: drbd, enable/disable data “barriers”</li>
</ul>
</dd>
</dl>
<p>Some parameters are bound to be disk-type specific (drbd, vs lvm, vs
files) or hypervisor specific (nic models for example), but for now they
will all live in the same structure. Each component is supposed to
validate only the parameters it knows about, and ganeti itself will make
sure that no “globally unknown” parameters are added, and that no
parameters have overridden meanings for different components.</p>
<p>The parameters will be kept, as for the BEPARAMS into a “default”
category, which will allow us to expand on by creating instance
“classes” in the future. Instance classes is not a feature we plan
implementing in 2.1, though.</p>
</div>
</div>
<div class="section" id="global-hypervisor-parameters">
<h4><a class="toc-backref" href="#id32">Global hypervisor parameters</a><a class="headerlink" href="#global-hypervisor-parameters" title="Permalink to this headline">¶</a></h4>
<div class="section" id="id11">
<h5>Current State and shortcomings<a class="headerlink" href="#id11" title="Permalink to this headline">¶</a></h5>
<p>Currently all hypervisor parameters are modifiable both globally
(cluster level) and at instance level. However, there is no other
framework to held hypervisor-specific parameters, so if we want to add
a new class of hypervisor parameters that only makes sense on a global
level, we have to change the hvparams framework.</p>
</div>
<div class="section" id="id12">
<h5>Proposed changes<a class="headerlink" href="#id12" title="Permalink to this headline">¶</a></h5>
<p>We add a new (global, not per-hypervisor) list of parameters which are
not changeable on a per-instance level. The create, modify and query
instance operations are changed to not allow/show these parameters.</p>
<p>Furthermore, to allow transition of parameters to the global list, and
to allow cleanup of inadverdently-customised parameters, the
<code class="docutils literal"><span class="pre">UpgradeConfig()</span></code> method of instances will drop any such parameters
from their list of hvparams, such that a restart of the master daemon
is all that is needed for cleaning these up.</p>
<p>Also, the framework is simple enough that if we need to replicate it
at beparams level we can do so easily.</p>
</div>
</div>
<div class="section" id="non-bridged-instances-support">
<h4><a class="toc-backref" href="#id33">Non bridged instances support</a><a class="headerlink" href="#non-bridged-instances-support" title="Permalink to this headline">¶</a></h4>
<div class="section" id="id13">
<h5>Current State and shortcomings<a class="headerlink" href="#id13" title="Permalink to this headline">¶</a></h5>
<p>Currently each instance NIC must be connected to a bridge, and if the
bridge is not specified the default cluster one is used. This makes it
impossible to use the vif-route xen network scripts, or other
alternative mechanisms that don’t need a bridge to work.</p>
</div>
<div class="section" id="id14">
<h5>Proposed changes<a class="headerlink" href="#id14" title="Permalink to this headline">¶</a></h5>
<p>The new “mode” network parameter will distinguish between bridged
interfaces and routed ones.</p>
<p>When mode is “bridge” the “link” parameter will contain the bridge the
instance should be connected to, effectively making things as today. The
value has been migrated from a nic field to a parameter to allow for an
easier manipulation of the cluster default.</p>
<p>When mode is “route” the ip field of the interface will become
mandatory, to allow for a route to be set. In the future we may want
also to accept multiple IPs or IP/mask values for this purpose. We will
evaluate possible meanings of the link parameter to signify a routing
table to be used, which would allow for insulation between instance
groups (as today happens for different bridges).</p>
<p>For now we won’t add a parameter to specify which network script gets
called for which instance, so in a mixed cluster the network script must
be able to handle both cases. The default kvm vif script will be changed
to do so. (Xen doesn’t have a ganeti provided script, so nothing will be
done for that hypervisor)</p>
</div>
</div>
<div class="section" id="introducing-persistent-uuids">
<h4><a class="toc-backref" href="#id34">Introducing persistent UUIDs</a><a class="headerlink" href="#introducing-persistent-uuids" title="Permalink to this headline">¶</a></h4>
<div class="section" id="id15">
<h5>Current state and shortcomings<a class="headerlink" href="#id15" title="Permalink to this headline">¶</a></h5>
<p>Some objects in the Ganeti configurations are tracked by their name
while also supporting renames. This creates an extra difficulty,
because neither Ganeti nor external management tools can then track
the actual entity, and due to the name change it behaves like a new
one.</p>
</div>
<div class="section" id="proposed-changes-part-1">
<h5>Proposed changes part 1<a class="headerlink" href="#proposed-changes-part-1" title="Permalink to this headline">¶</a></h5>
<p>We will change Ganeti to use UUIDs for entity tracking, but in a
staggered way. In 2.1, we will simply add an “uuid” attribute to each
of the instances, nodes and cluster itself. This will be reported on
instance creation for nodes, and on node adds for the nodes. It will
be of course avaiblable for querying via the OpNodeQuery/Instance and
cluster information, and via RAPI as well.</p>
<p>Note that Ganeti will not provide any way to change this attribute.</p>
<p>Upgrading from Ganeti 2.0 will automatically add an ‘uuid’ attribute
to all entities missing it.</p>
</div>
<div class="section" id="proposed-changes-part-2">
<h5>Proposed changes part 2<a class="headerlink" href="#proposed-changes-part-2" title="Permalink to this headline">¶</a></h5>
<p>In the next release (e.g. 2.2), the tracking of objects will change
from the name to the UUID internally, and externally Ganeti will
accept both forms of identification; e.g. an RAPI call would be made
either against <code class="docutils literal"><span class="pre">/2/instances/foo.bar</span></code> or against
<code class="docutils literal"><span class="pre">/2/instances/bb3b2e42…</span></code>. Since an FQDN must have at least a dot,
and dots are not valid characters in UUIDs, we will not have namespace
issues.</p>
<p>Another change here is that node identification (during cluster
operations/queries like master startup, “am I the master?” and
similar) could be done via UUIDs which is more stable than the current
hostname-based scheme.</p>
<p>Internal tracking refers to the way the configuration is stored; a
DRBD disk of an instance refers to the node name (so that IPs can be
changed easily), but this is still a problem for name changes; thus
these will be changed to point to the node UUID to ease renames.</p>
<p>The advantages of this change (after the second round of changes), is
that node rename becomes trivial, whereas today node rename would
require a complete lock of all instances.</p>
</div>
</div>
<div class="section" id="automated-disk-repairs-infrastructure">
<h4><a class="toc-backref" href="#id35">Automated disk repairs infrastructure</a><a class="headerlink" href="#automated-disk-repairs-infrastructure" title="Permalink to this headline">¶</a></h4>
<p>Replacing defective disks in an automated fashion is quite difficult
with the current version of Ganeti. These changes will introduce
additional functionality and interfaces to simplify automating disk
replacements on a Ganeti node.</p>
<div class="section" id="fix-node-volume-group">
<h5>Fix node volume group<a class="headerlink" href="#fix-node-volume-group" title="Permalink to this headline">¶</a></h5>
<p>This is the most difficult addition, as it can lead to dataloss if it’s
not properly safeguarded.</p>
<p>The operation must be done only when all the other nodes that have
instances in common with the target node are fine, i.e. this is the only
node with problems, and also we have to double-check that all instances
on this node have at least a good copy of the data.</p>
<p>This might mean that we have to enhance the GetMirrorStatus calls, and
introduce and a smarter version that can tell us more about the status
of an instance.</p>
</div>
<div class="section" id="stop-allocation-on-a-given-pv">
<h5>Stop allocation on a given PV<a class="headerlink" href="#stop-allocation-on-a-given-pv" title="Permalink to this headline">¶</a></h5>
<p>This is somewhat simple. First we need a “list PVs” opcode (and its
associated logical unit) and then a set PV status opcode/LU. These in
combination should allow both checking and changing the disk/PV status.</p>
</div>
<div class="section" id="instance-disk-status">
<h5>Instance disk status<a class="headerlink" href="#instance-disk-status" title="Permalink to this headline">¶</a></h5>
<p>This new opcode or opcode change must list the instance-disk-index and
node combinations of the instance together with their status. This will
allow determining what part of the instance is broken (if any).</p>
</div>
<div class="section" id="repair-instance">
<h5>Repair instance<a class="headerlink" href="#repair-instance" title="Permalink to this headline">¶</a></h5>
<p>This new opcode/LU/RAPI call will run <code class="docutils literal"><span class="pre">replace-disks</span> <span class="pre">-p</span></code> as needed, in
order to fix the instance status. It only affects primary instances;
secondaries can just be moved away.</p>
</div>
<div class="section" id="migrate-node">
<h5>Migrate node<a class="headerlink" href="#migrate-node" title="Permalink to this headline">¶</a></h5>
<p>This new opcode/LU/RAPI call will take over the current <code class="docutils literal"><span class="pre">gnt-node</span>
<span class="pre">migrate</span></code> code and run migrate for all instances on the node.</p>
</div>
<div class="section" id="evacuate-node">
<h5>Evacuate node<a class="headerlink" href="#evacuate-node" title="Permalink to this headline">¶</a></h5>
<p>This new opcode/LU/RAPI call will take over the current <code class="docutils literal"><span class="pre">gnt-node</span>
<span class="pre">evacuate</span></code> code and run replace-secondary with an iallocator script for
all instances on the node.</p>
</div>
</div>
<div class="section" id="user-id-pool">
<h4><a class="toc-backref" href="#id36">User-id pool</a><a class="headerlink" href="#user-id-pool" title="Permalink to this headline">¶</a></h4>
<p>In order to allow running different processes under unique user-ids
on a node, we introduce the user-id pool concept.</p>
<p>The user-id pool is a cluster-wide configuration parameter.
It is a list of user-ids and/or user-id ranges that are reserved
for running Ganeti processes (including KVM instances).
The code guarantees that on a given node a given user-id is only
handed out if there is no other process running with that user-id.</p>
<p>Please note, that this can only be guaranteed if all processes in
the system - that run under a user-id belonging to the pool - are
started by reserving a user-id first. That can be accomplished
either by using the RequestUnusedUid() function to get an unused
user-id or by implementing the same locking mechanism.</p>
<div class="section" id="implementation">
<h5>Implementation<a class="headerlink" href="#implementation" title="Permalink to this headline">¶</a></h5>
<p>The functions that are specific to the user-id pool feature are located
in a separate module: <code class="docutils literal"><span class="pre">lib/uidpool.py</span></code>.</p>
<div class="section" id="storage">
<h6>Storage<a class="headerlink" href="#storage" title="Permalink to this headline">¶</a></h6>
<p>The user-id pool is a single cluster parameter. It is stored in the
<em>Cluster</em> object under the <code class="docutils literal"><span class="pre">uid_pool</span></code> name as a list of integer
tuples. These tuples represent the boundaries of user-id ranges.
For single user-ids, the boundaries are equal.</p>
<p>The internal user-id pool representation is converted into a
string: a newline separated list of user-ids or user-id ranges.
This string representation is distributed to all the nodes via the
<em>ssconf</em> mechanism. This means that the user-id pool can be
accessed in a read-only way on any node without consulting the master
node or master candidate nodes.</p>
</div>
<div class="section" id="initial-value">
<h6>Initial value<a class="headerlink" href="#initial-value" title="Permalink to this headline">¶</a></h6>
<p>The value of the user-id pool cluster parameter can be initialized
at cluster initialization time using the</p>
<p><code class="docutils literal"><span class="pre">gnt-cluster</span> <span class="pre">init</span> <span class="pre">--uid-pool</span> <span class="pre"><uid-pool</span> <span class="pre">definition></span> <span class="pre">...</span></code></p>
<p>command.</p>
<p>As there is no sensible default value for the user-id pool parameter,
it is initialized to an empty list if no <code class="docutils literal"><span class="pre">--uid-pool</span></code> option is
supplied at cluster init time.</p>
<p>If the user-id pool is empty, the user-id pool feature is considered
to be disabled.</p>
</div>
<div class="section" id="manipulation">
<h6>Manipulation<a class="headerlink" href="#manipulation" title="Permalink to this headline">¶</a></h6>
<p>The user-id pool cluster parameter can be modified from the
command-line with the following commands:</p>
<ul class="simple">
<li><code class="docutils literal"><span class="pre">gnt-cluster</span> <span class="pre">modify</span> <span class="pre">--uid-pool</span> <span class="pre"><uid-pool</span> <span class="pre">definition></span></code></li>
<li><code class="docutils literal"><span class="pre">gnt-cluster</span> <span class="pre">modify</span> <span class="pre">--add-uids</span> <span class="pre"><uid-pool</span> <span class="pre">definition></span></code></li>
<li><code class="docutils literal"><span class="pre">gnt-cluster</span> <span class="pre">modify</span> <span class="pre">--remove-uids</span> <span class="pre"><uid-pool</span> <span class="pre">definition></span></code></li>
</ul>
<p>The <code class="docutils literal"><span class="pre">--uid-pool</span></code> option overwrites the current setting with the
supplied <code class="docutils literal"><span class="pre"><uid-pool</span> <span class="pre">definition></span></code>, while
<code class="docutils literal"><span class="pre">--add-uids</span></code>/<code class="docutils literal"><span class="pre">--remove-uids</span></code> adds/removes the listed uids
or uid-ranges from the pool.</p>
<p>The <code class="docutils literal"><span class="pre"><uid-pool</span> <span class="pre">definition></span></code> should be a comma-separated list of
user-ids or user-id ranges. A range should be defined by a lower and
a higher boundary. The boundaries should be separated with a dash.
The boundaries are inclusive.</p>
<p>The <code class="docutils literal"><span class="pre"><uid-pool</span> <span class="pre">definition></span></code> is parsed into the internal
representation, sanity-checked and stored in the <code class="docutils literal"><span class="pre">uid_pool</span></code>
attribute of the <em>Cluster</em> object.</p>
<p>It is also immediately converted into a string (formatted in the
input format) and distributed to all nodes via the <em>ssconf</em> mechanism.</p>
</div>
<div class="section" id="inspection">
<h6>Inspection<a class="headerlink" href="#inspection" title="Permalink to this headline">¶</a></h6>
<p>The current value of the user-id pool cluster parameter is printed
by the <code class="docutils literal"><span class="pre">gnt-cluster</span> <span class="pre">info</span></code> command.</p>
<p>The output format is accepted by the <code class="docutils literal"><span class="pre">gnt-cluster</span> <span class="pre">modify</span> <span class="pre">--uid-pool</span></code>
command.</p>
</div>
<div class="section" id="locking">
<h6>Locking<a class="headerlink" href="#locking" title="Permalink to this headline">¶</a></h6>
<p>The <code class="docutils literal"><span class="pre">uidpool.py</span></code> module provides a function (<code class="docutils literal"><span class="pre">RequestUnusedUid</span></code>)
for requesting an unused user-id from the pool.</p>
<p>This will try to find a random user-id that is not currently in use.
The algorithm is the following:</p>
<ol class="arabic simple">
<li>Randomize the list of user-ids in the user-id pool</li>
<li>Iterate over this randomized UID list</li>
<li>Create a lock file (it doesn’t matter if it already exists)</li>
<li>Acquire an exclusive POSIX lock on the file, to provide mutual
exclusion for the following non-atomic operations</li>
<li>Check if there is a process in the system with the given UID</li>
<li>If there isn’t, return the UID, otherwise unlock the file and
continue the iteration over the user-ids</li>
</ol>
<p>The user can than start a new process with this user-id.
Once a process is successfully started, the exclusive POSIX lock can
be released, but the lock file will remain in the filesystem.
The presence of such a lock file means that the given user-id is most
probably in use. The lack of a uid lock file does not guarantee that
there are no processes with that user-id.</p>
<p>After acquiring the exclusive POSIX lock, <code class="docutils literal"><span class="pre">RequestUnusedUid</span></code>
always performs a check to see if there is a process running with the
given uid.</p>
<p>A user-id can be returned to the pool, by calling the
<code class="docutils literal"><span class="pre">ReleaseUid</span></code> function. This will remove the corresponding lock file.
Note, that it doesn’t check if there is any process still running
with that user-id. The removal of the lock file only means that there
are most probably no processes with the given user-id. This helps
in speeding up the process of finding a user-id that is guaranteed to
be unused.</p>
<p>There is a convenience function, called <code class="docutils literal"><span class="pre">ExecWithUnusedUid</span></code> that
wraps the execution of a function (or any callable) that requires a
unique user-id. <code class="docutils literal"><span class="pre">ExecWithUnusedUid</span></code> takes care of requesting an
unused user-id and unlocking the lock file. It also automatically
returns the user-id to the pool if the callable raises an exception.</p>
</div>
</div>
<div class="section" id="code-examples">
<h5>Code examples<a class="headerlink" href="#code-examples" title="Permalink to this headline">¶</a></h5>
<p>Requesting a user-id from the pool:</p>
<div class="highlight-default"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">ganeti</span> <span class="k">import</span> <span class="n">ssconf</span>
<span class="kn">from</span> <span class="nn">ganeti</span> <span class="k">import</span> <span class="n">uidpool</span>
<span class="c1"># Get list of all user-ids in the uid-pool from ssconf</span>
<span class="n">ss</span> <span class="o">=</span> <span class="n">ssconf</span><span class="o">.</span><span class="n">SimpleStore</span><span class="p">()</span>
<span class="n">uid_pool</span> <span class="o">=</span> <span class="n">uidpool</span><span class="o">.</span><span class="n">ParseUidPool</span><span class="p">(</span><span class="n">ss</span><span class="o">.</span><span class="n">GetUidPool</span><span class="p">(),</span> <span class="n">separator</span><span class="o">=</span><span class="s2">"</span><span class="se">\n</span><span class="s2">"</span><span class="p">)</span>
<span class="n">all_uids</span> <span class="o">=</span> <span class="nb">set</span><span class="p">(</span><span class="n">uidpool</span><span class="o">.</span><span class="n">ExpandUidPool</span><span class="p">(</span><span class="n">uid_pool</span><span class="p">))</span>
<span class="n">uid</span> <span class="o">=</span> <span class="n">uidpool</span><span class="o">.</span><span class="n">RequestUnusedUid</span><span class="p">(</span><span class="n">all_uids</span><span class="p">)</span>
<span class="k">try</span><span class="p">:</span>
<span class="o"><</span><span class="n">start</span> <span class="n">a</span> <span class="n">process</span> <span class="k">with</span> <span class="n">the</span> <span class="n">UID</span><span class="o">></span>
<span class="c1"># Once the process is started, we can release the file lock</span>
<span class="n">uid</span><span class="o">.</span><span class="n">Unlock</span><span class="p">()</span>
<span class="k">except</span> <span class="o">...</span><span class="p">,</span> <span class="n">err</span><span class="p">:</span>
<span class="c1"># Return the UID to the pool</span>
<span class="n">uidpool</span><span class="o">.</span><span class="n">ReleaseUid</span><span class="p">(</span><span class="n">uid</span><span class="p">)</span>
</pre></div>
</div>
<p>Releasing a user-id:</p>
<div class="highlight-default"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">ganeti</span> <span class="k">import</span> <span class="n">uidpool</span>
<span class="n">uid</span> <span class="o">=</span> <span class="o"><</span><span class="n">get</span> <span class="n">the</span> <span class="n">UID</span> <span class="n">the</span> <span class="n">process</span> <span class="ow">is</span> <span class="n">running</span> <span class="n">under</span><span class="o">></span>
<span class="o"><</span><span class="n">stop</span> <span class="n">the</span> <span class="n">process</span><span class="o">></span>
<span class="n">uidpool</span><span class="o">.</span><span class="n">ReleaseUid</span><span class="p">(</span><span class="n">uid</span><span class="p">)</span>
</pre></div>
</div>
</div>
</div>
</div>
<div class="section" id="external-interface-changes">
<h3><a class="toc-backref" href="#id37">External interface changes</a><a class="headerlink" href="#external-interface-changes" title="Permalink to this headline">¶</a></h3>
<div class="section" id="os-api">
<h4><a class="toc-backref" href="#id38">OS API</a><a class="headerlink" href="#os-api" title="Permalink to this headline">¶</a></h4>
<p>The OS API of Ganeti 2.0 has been built with extensibility in mind.
Since we pass everything as environment variables it’s a lot easier to
send new information to the OSes without breaking retrocompatibility.
This section of the design outlines the proposed extensions to the API
and their implementation.</p>
<div class="section" id="api-version-compatibility-handling">
<h5>API Version Compatibility Handling<a class="headerlink" href="#api-version-compatibility-handling" title="Permalink to this headline">¶</a></h5>
<p>In 2.1 there will be a new OS API version (eg. 15), which should be
mostly compatible with api 10, except for some new added variables.
Since it’s easy not to pass some variables we’ll be able to handle
Ganeti 2.0 OSes by just filtering out the newly added piece of
information. We will still encourage OSes to declare support for the new
API after checking that the new variables don’t provide any conflict for
them, and we will drop api 10 support after ganeti 2.1 has released.</p>
</div>
<div class="section" id="new-environment-variables">
<h5>New Environment variables<a class="headerlink" href="#new-environment-variables" title="Permalink to this headline">¶</a></h5>
<p>Some variables have never been added to the OS api but would definitely
be useful for the OSes. We plan to add an INSTANCE_HYPERVISOR variable
to allow the OS to make changes relevant to the virtualization the
instance is going to use. Since this field is immutable for each
instance, the os can tight the install without caring of making sure the
instance can run under any virtualization technology.</p>
<p>We also want the OS to know the particular hypervisor parameters, to be
able to customize the install even more. Since the parameters can
change, though, we will pass them only as an “FYI”: if an OS ties some
instance functionality to the value of a particular hypervisor parameter
manual changes or a reinstall may be needed to adapt the instance to the
new environment. This is not a regression as of today, because even if
the OSes are left blind about this information, sometimes they still
need to make compromises and cannot satisfy all possible parameter
values.</p>
</div>
<div class="section" id="os-variants">
<h5>OS Variants<a class="headerlink" href="#os-variants" title="Permalink to this headline">¶</a></h5>
<p>Currently we are assisting to some degree of “os proliferation” just to
change a simple installation behavior. This means that the same OS gets
installed on the cluster multiple times, with different names, to
customize just one installation behavior. Usually such OSes try to share
as much as possible through symlinks, but this still causes
complications on the user side, especially when multiple parameters must
be cross-matched.</p>
<p>For example today if you want to install debian etch, lenny or squeeze
you probably need to install the debootstrap OS multiple times, changing
its configuration file, and calling it debootstrap-etch,
debootstrap-lenny or debootstrap-squeeze. Furthermore if you have for
example a “server” and a “development” environment which installs
different packages/configuration files and must be available for all
installs you’ll probably end up with deboostrap-etch-server,
debootstrap-etch-dev, debootrap-lenny-server, debootstrap-lenny-dev,
etc. Crossing more than two parameters quickly becomes not manageable.</p>
<p>In order to avoid this we plan to make OSes more customizable, by
allowing each OS to declare a list of variants which can be used to
customize it. The variants list is mandatory and must be written, one
variant per line, in the new “variants.list” file inside the main os
dir. At least one supported variant must be supported. When choosing the
OS exactly one variant will have to be specified, and will be encoded in
the os name as <OS-name>+<variant>. As for today it will be possible to
change an instance’s OS at creation or install time.</p>
<p>The 2.1 OS list will be the combination of each OS, plus its supported
variants. This will cause the name name proliferation to remain, but at
least the internal OS code will be simplified to just parsing the passed
variant, without the need for symlinks or code duplication.</p>
<p>Also we expect the OSes to declare only “interesting” variants, but to
accept some non-declared ones which a user will be able to pass in by
overriding the checks ganeti does. This will be useful for allowing some
variations to be used without polluting the OS list (per-OS
documentation should list all supported variants). If a variant which is
not internally supported is forced through, the OS scripts should abort.</p>
<p>In the future (post 2.1) we may want to move to full fledged parameters
all orthogonal to each other (for example “architecture” (i386, amd64),
“suite” (lenny, squeeze, …), etc). (As opposed to the variant, which
is a single parameter, and you need a different variant for all the set
of combinations you want to support). In this case we envision the
variants to be moved inside of Ganeti and be associated with lists
parameter->values associations, which will then be passed to the OS.</p>
</div>
</div>
<div class="section" id="iallocator-changes">
<h4><a class="toc-backref" href="#id39">IAllocator changes</a><a class="headerlink" href="#iallocator-changes" title="Permalink to this headline">¶</a></h4>
<div class="section" id="id16">
<h5>Current State and shortcomings<a class="headerlink" href="#id16" title="Permalink to this headline">¶</a></h5>
<p>The iallocator interface allows creation of instances without manually
specifying nodes, but instead by specifying plugins which will do the
required computations and produce a valid node list.</p>
<p>However, the interface is quite akward to use:</p>
<ul class="simple">
<li>one cannot set a ‘default’ iallocator script</li>
<li>one cannot use it to easily test if allocation would succeed</li>
<li>some new functionality, such as rebalancing clusters and calculating
capacity estimates is needed</li>
</ul>
</div>
<div class="section" id="id17">
<h5>Proposed changes<a class="headerlink" href="#id17" title="Permalink to this headline">¶</a></h5>
<p>There are two area of improvements proposed:</p>
<ul class="simple">
<li>improving the use of the current interface</li>
<li>extending the IAllocator API to cover more automation</li>
</ul>
<div class="section" id="default-iallocator-names">
<h6>Default iallocator names<a class="headerlink" href="#default-iallocator-names" title="Permalink to this headline">¶</a></h6>
<p>The cluster will hold, for each type of iallocator, a (possibly empty)
list of modules that will be used automatically.</p>
<p>If the list is empty, the behaviour will remain the same.</p>
<p>If the list has one entry, then ganeti will behave as if
‘–iallocator’ was specifyed on the command line. I.e. use this
allocator by default. If the user however passed nodes, those will be
used in preference.</p>
<p>If the list has multiple entries, they will be tried in order until
one gives a successful answer.</p>
</div>
<div class="section" id="dry-run-allocation">
<h6>Dry-run allocation<a class="headerlink" href="#dry-run-allocation" title="Permalink to this headline">¶</a></h6>
<p>The create instance LU will get a new ‘dry-run’ option that will just
simulate the placement, and return the chosen node-lists after running
all the usual checks.</p>
</div>
<div class="section" id="cluster-balancing">
<h6>Cluster balancing<a class="headerlink" href="#cluster-balancing" title="Permalink to this headline">¶</a></h6>
<p>Instance add/removals/moves can create a situation where load on the
nodes is not spread equally. For this, a new iallocator mode will be
implemented called <code class="docutils literal"><span class="pre">balance</span></code> in which the plugin, given the current
cluster state, and a maximum number of operations, will need to
compute the instance relocations needed in order to achieve a “better”
(for whatever the script believes it’s better) cluster.</p>
</div>
<div class="section" id="cluster-capacity-calculation">
<h6>Cluster capacity calculation<a class="headerlink" href="#cluster-capacity-calculation" title="Permalink to this headline">¶</a></h6>
<p>In this mode, called <code class="docutils literal"><span class="pre">capacity</span></code>, given an instance specification and
the current cluster state (similar to the <code class="docutils literal"><span class="pre">allocate</span></code> mode), the
plugin needs to return:</p>
<ul class="simple">
<li>how many instances can be allocated on the cluster with that
specification</li>
<li>on which nodes these will be allocated (in order)</li>
</ul>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
<div class="sphinxsidebar" role="navigation" aria-label="main navigation">
<div class="sphinxsidebarwrapper">
<h3><a href="index.html">Table Of Contents</a></h3>
<ul>
<li><a class="reference internal" href="#">Ganeti 2.1 design</a><ul>
<li><a class="reference internal" href="#objective">Objective</a></li>
<li><a class="reference internal" href="#detailed-design">Detailed design</a><ul>
<li><a class="reference internal" href="#core-changes">Core changes</a><ul>
<li><a class="reference internal" href="#storage-units-modelling">Storage units modelling</a></li>
<li><a class="reference internal" href="#locking-improvements">Locking improvements</a><ul>
<li><a class="reference internal" href="#current-state-and-shortcomings">Current State and shortcomings</a></li>
<li><a class="reference internal" href="#proposed-changes">Proposed changes</a><ul>
<li><a class="reference internal" href="#non-blocking-lock-acquiring">Non-blocking lock acquiring</a></li>
<li><a class="reference internal" href="#retry-acquiring-locks">Retry acquiring locks</a></li>
</ul>
</li>
<li><a class="reference internal" href="#other-solutions-discussed">Other solutions discussed</a></li>
<li><a class="reference internal" href="#implementation-details">Implementation details</a><ul>
<li><a class="reference internal" href="#sharedlock-redesign"><code class="docutils literal"><span class="pre">SharedLock</span></code> redesign</a><ul>
<li><a class="reference internal" href="#acquire">Acquire</a></li>
<li><a class="reference internal" href="#release">Release</a></li>
<li><a class="reference internal" href="#delete">Delete</a></li>
</ul>
</li>
<li><a class="reference internal" href="#condition">Condition</a></li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#node-daemon-availability">Node daemon availability</a><ul>
<li><a class="reference internal" href="#id1">Current State and shortcomings</a></li>
<li><a class="reference internal" href="#id2">Proposed changes</a></li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#new-features">New Features</a><ul>
<li><a class="reference internal" href="#automated-ganeti-cluster-merger">Automated Ganeti Cluster Merger</a><ul>
<li><a class="reference internal" href="#current-situation">Current situation</a></li>
<li><a class="reference internal" href="#initial-contact">Initial contact</a></li>
<li><a class="reference internal" href="#cluster-merge">Cluster merge</a></li>
<li><a class="reference internal" href="#rollback">Rollback</a></li>
<li><a class="reference internal" href="#verification">Verification</a></li>
<li><a class="reference internal" href="#appendix">Appendix</a><ul>
<li><a class="reference internal" href="#cluster-merge-py">cluster-merge.py</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#feature-changes">Feature changes</a><ul>
<li><a class="reference internal" href="#ganeti-confd">Ganeti Confd</a><ul>
<li><a class="reference internal" href="#id3">Current State and shortcomings</a></li>
<li><a class="reference internal" href="#id4">Proposed changes</a><ul>
<li><a class="reference internal" href="#wire-protocol">Wire protocol</a></li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#redistribute-config">Redistribute Config</a><ul>
<li><a class="reference internal" href="#id5">Current State and shortcomings</a></li>
<li><a class="reference internal" href="#id6">Proposed changes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#vnc-console-password">VNC Console Password</a><ul>
<li><a class="reference internal" href="#id7">Current State and shortcomings</a></li>
<li><a class="reference internal" href="#id8">Proposed changes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#disk-net-parameters">Disk/Net parameters</a><ul>
<li><a class="reference internal" href="#id9">Current State and shortcomings</a></li>
<li><a class="reference internal" href="#id10">Proposed changes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#global-hypervisor-parameters">Global hypervisor parameters</a><ul>
<li><a class="reference internal" href="#id11">Current State and shortcomings</a></li>
<li><a class="reference internal" href="#id12">Proposed changes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#non-bridged-instances-support">Non bridged instances support</a><ul>
<li><a class="reference internal" href="#id13">Current State and shortcomings</a></li>
<li><a class="reference internal" href="#id14">Proposed changes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#introducing-persistent-uuids">Introducing persistent UUIDs</a><ul>
<li><a class="reference internal" href="#id15">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#proposed-changes-part-1">Proposed changes part 1</a></li>
<li><a class="reference internal" href="#proposed-changes-part-2">Proposed changes part 2</a></li>
</ul>
</li>
<li><a class="reference internal" href="#automated-disk-repairs-infrastructure">Automated disk repairs infrastructure</a><ul>
<li><a class="reference internal" href="#fix-node-volume-group">Fix node volume group</a></li>
<li><a class="reference internal" href="#stop-allocation-on-a-given-pv">Stop allocation on a given PV</a></li>
<li><a class="reference internal" href="#instance-disk-status">Instance disk status</a></li>
<li><a class="reference internal" href="#repair-instance">Repair instance</a></li>
<li><a class="reference internal" href="#migrate-node">Migrate node</a></li>
<li><a class="reference internal" href="#evacuate-node">Evacuate node</a></li>
</ul>
</li>
<li><a class="reference internal" href="#user-id-pool">User-id pool</a><ul>
<li><a class="reference internal" href="#implementation">Implementation</a><ul>
<li><a class="reference internal" href="#storage">Storage</a></li>
<li><a class="reference internal" href="#initial-value">Initial value</a></li>
<li><a class="reference internal" href="#manipulation">Manipulation</a></li>
<li><a class="reference internal" href="#inspection">Inspection</a></li>
<li><a class="reference internal" href="#locking">Locking</a></li>
</ul>
</li>
<li><a class="reference internal" href="#code-examples">Code examples</a></li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#external-interface-changes">External interface changes</a><ul>
<li><a class="reference internal" href="#os-api">OS API</a><ul>
<li><a class="reference internal" href="#api-version-compatibility-handling">API Version Compatibility Handling</a></li>
<li><a class="reference internal" href="#new-environment-variables">New Environment variables</a></li>
<li><a class="reference internal" href="#os-variants">OS Variants</a></li>
</ul>
</li>
<li><a class="reference internal" href="#iallocator-changes">IAllocator changes</a><ul>
<li><a class="reference internal" href="#id16">Current State and shortcomings</a></li>
<li><a class="reference internal" href="#id17">Proposed changes</a><ul>
<li><a class="reference internal" href="#default-iallocator-names">Default iallocator names</a></li>
<li><a class="reference internal" href="#dry-run-allocation">Dry-run allocation</a></li>
<li><a class="reference internal" href="#cluster-balancing">Cluster balancing</a></li>
<li><a class="reference internal" href="#cluster-capacity-calculation">Cluster capacity calculation</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
<h4>Previous topic</h4>
<p class="topless"><a href="design-2.0.html"
title="previous chapter">Ganeti 2.0 design</a></p>
<h4>Next topic</h4>
<p class="topless"><a href="design-2.2.html"
title="next chapter">Ganeti 2.2 design</a></p>
<div role="note" aria-label="source link">
<h3>This Page</h3>
<ul class="this-page-menu">
<li><a href="_sources/design-2.1.rst.txt"
rel="nofollow">Show Source</a></li>
</ul>
</div>
<div id="searchbox" style="display: none" role="search">
<h3>Quick search</h3>
<form class="search" action="search.html" method="get">
<div><input type="text" name="q" /></div>
<div><input type="submit" value="Go" /></div>
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
<script type="text/javascript">$('#searchbox').show(0);</script>
</div>
</div>
<div class="clearer"></div>
</div>
<div class="related" role="navigation" aria-label="related navigation">
<h3>Navigation</h3>
<ul>
<li class="right" style="margin-right: 10px">
<a href="design-2.2.html" title="Ganeti 2.2 design"
>next</a></li>
<li class="right" >
<a href="design-2.0.html" title="Ganeti 2.0 design"
>previous</a> |</li>
<li class="nav-item nav-item-0"><a href="index.html">Ganeti 2.16.0~rc2 documentation</a> »</li>
</ul>
</div>
<div class="footer" role="contentinfo">
© Copyright 2018, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015 Google Inc..
Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.6.7.
</div>
</body>
</html>
|