This file is indexed.

/usr/share/doc/ganeti/html/design-2.2.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
<!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 2.2 design &#8212; 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 2.3 design" href="design-2.3.html" />
    <link rel="prev" title="Ganeti 2.1 design" href="design-2.1.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-2.3.html" title="Ganeti 2.3 design"
             accesskey="N">next</a></li>
        <li class="right" >
          <a href="design-2.1.html" title="Ganeti 2.1 design"
             accesskey="P">previous</a> |</li>
        <li class="nav-item nav-item-0"><a href="index.html">Ganeti 2.16.0~rc2 documentation</a> &#187;</li> 
      </ul>
    </div>  

    <div class="document">
      <div class="documentwrapper">
        <div class="bodywrapper">
          <div class="body" role="main">
            
  <div class="section" id="ganeti-2-2-design">
<h1><a class="toc-backref" href="#id10">Ganeti 2.2 design</a><a class="headerlink" href="#ganeti-2-2-design" title="Permalink to this headline"></a></h1>
<p>This document describes the major changes in Ganeti 2.2 compared to
the 2.1 version.</p>
<p>The 2.2 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 2.1, 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-2-2-design" id="id10">Ganeti 2.2 design</a><ul>
<li><a class="reference internal" href="#core-changes" id="id11">Core changes</a><ul>
<li><a class="reference internal" href="#master-daemon-scaling-improvements" id="id12">Master Daemon Scaling improvements</a><ul>
<li><a class="reference internal" href="#current-state-and-shortcomings" id="id13">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#proposed-changes" id="id14">Proposed changes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#inter-cluster-instance-moves" id="id15">Inter-cluster instance moves</a><ul>
<li><a class="reference internal" href="#id3" id="id16">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#id4" id="id17">Proposed changes</a></li>
</ul>
</li>
<li><a class="reference internal" href="#privilege-separation" id="id18">Privilege separation</a><ul>
<li><a class="reference internal" href="#id5" id="id19">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#implementation" id="id20">Implementation</a></li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#feature-changes" id="id21">Feature changes</a><ul>
<li><a class="reference internal" href="#kvm-security" id="id22">KVM Security</a><ul>
<li><a class="reference internal" href="#id6" id="id23">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#id7" id="id24">Proposed changes</a></li>
<li><a class="reference internal" href="#implementation-plan" id="id25">Implementation plan</a></li>
</ul>
</li>
<li><a class="reference internal" href="#new-os-states" id="id26">New OS states</a><ul>
<li><a class="reference internal" href="#id8" id="id27">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#id9" id="id28">Proposed changes</a></li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#external-interface-changes" id="id29">External interface changes</a><ul>
<li><a class="reference internal" href="#os-api" id="id30">OS API</a><ul>
<li><a class="reference internal" href="#os-version" id="id31">OS version</a></li>
<li><a class="reference internal" href="#parameters" id="id32">Parameters</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</div>
<p>As for 2.1 we divide the 2.2 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 (e.g. command line, OS API, hooks, …)</li>
</ul>
<div class="section" id="core-changes">
<h2><a class="toc-backref" href="#id11">Core changes</a><a class="headerlink" href="#core-changes" title="Permalink to this headline"></a></h2>
<div class="section" id="master-daemon-scaling-improvements">
<h3><a class="toc-backref" href="#id12">Master Daemon Scaling improvements</a><a class="headerlink" href="#master-daemon-scaling-improvements" title="Permalink to this headline"></a></h3>
<div class="section" id="current-state-and-shortcomings">
<h4><a class="toc-backref" href="#id13">Current state and shortcomings</a><a class="headerlink" href="#current-state-and-shortcomings" title="Permalink to this headline"></a></h4>
<p>Currently the Ganeti master daemon is based on four sets of threads:</p>
<ul class="simple">
<li>The main thread (1 thread) just accepts connections on the master
socket</li>
<li>The client worker pool (16 threads) handles those connections,
one thread per connected socket, parses luxi requests, and sends data
back to the clients</li>
<li>The job queue worker pool (25 threads) executes the actual jobs
submitted by the clients</li>
<li>The rpc worker pool (10 threads) interacts with the nodes via
http-based-rpc</li>
</ul>
<p>This means that every masterd currently runs 52 threads to do its job.
Being able to reduce the number of thread sets would make the master’s
architecture a lot simpler. Moreover having less threads can help
decrease lock contention, log pollution and memory usage.
Also, with the current architecture, masterd suffers from quite a few
scalability issues:</p>
<div class="section" id="core-daemon-connection-handling">
<h5>Core daemon connection handling<a class="headerlink" href="#core-daemon-connection-handling" title="Permalink to this headline"></a></h5>
<p>Since the 16 client worker threads handle one connection each, it’s very
easy to exhaust them, by just connecting to masterd 16 times and not
sending any data. While we could perhaps make those pools resizable,
increasing the number of threads won’t help with lock contention nor
with better handling long running operations making sure the client is
informed that everything is proceeding, and doesn’t need to time out.</p>
</div>
<div class="section" id="wait-for-job-change">
<h5>Wait for job change<a class="headerlink" href="#wait-for-job-change" title="Permalink to this headline"></a></h5>
<p>The REQ_WAIT_FOR_JOB_CHANGE luxi operation makes the relevant client
thread block on its job for a relatively long time. This is another
easy way to exhaust the 16 client threads, and a place where clients
often time out, moreover this operation is negative for the job queue
lock contention (see below).</p>
</div>
<div class="section" id="job-queue-lock">
<h5>Job Queue lock<a class="headerlink" href="#job-queue-lock" title="Permalink to this headline"></a></h5>
<p>The job queue lock is quite heavily contended, and certain easily
reproducible workloads show that’s it’s very easy to put masterd in
trouble: for example running ~15 background instance reinstall jobs,
results in a master daemon that, even without having finished the
client worker threads, can’t answer simple job list requests, or
submit more jobs.</p>
<p>Currently the job queue lock is an exclusive non-fair lock insulating
the following job queue methods (called by the client workers).</p>
<blockquote>
<div><ul class="simple">
<li>AddNode</li>
<li>RemoveNode</li>
<li>SubmitJob</li>
<li>SubmitManyJobs</li>
<li>WaitForJobChanges</li>
<li>CancelJob</li>
<li>ArchiveJob</li>
<li>AutoArchiveJobs</li>
<li>QueryJobs</li>
<li>Shutdown</li>
</ul>
</div></blockquote>
<p>Moreover the job queue lock is acquired outside of the job queue in two
other classes:</p>
<blockquote>
<div><ul class="simple">
<li>jqueue._JobQueueWorker (in RunTask) before executing the opcode, after
finishing its executing and when handling an exception.</li>
<li>jqueue._OpExecCallbacks (in NotifyStart and Feedback) when the
processor (mcpu.Processor) is about to start working on the opcode
(after acquiring the necessary locks) and when any data is sent back
via the feedback function.</li>
</ul>
</div></blockquote>
<p>Of those the major critical points are:</p>
<blockquote>
<div><ul class="simple">
<li>Submit[Many]Job, QueryJobs, WaitForJobChanges, which can easily slow
down and block client threads up to making the respective clients
time out.</li>
<li>The code paths in NotifyStart, Feedback, and RunTask, which slow
down job processing between clients and otherwise non-related jobs.</li>
</ul>
</div></blockquote>
<p>To increase the pain:</p>
<blockquote>
<div><ul class="simple">
<li>WaitForJobChanges is a bad offender because it’s implemented with a
notified condition which awakes waiting threads, who then try to
acquire the global lock again</li>
<li>Many should-be-fast code paths are slowed down by replicating the
change to remote nodes, and thus waiting, with the lock held, on
remote rpcs to complete (starting, finishing, and submitting jobs)</li>
</ul>
</div></blockquote>
</div>
</div>
<div class="section" id="proposed-changes">
<h4><a class="toc-backref" href="#id14">Proposed changes</a><a class="headerlink" href="#proposed-changes" title="Permalink to this headline"></a></h4>
<p>In order to be able to interact with the master daemon even when it’s
under heavy load, and  to make it simpler to add core functionality
(such as an asynchronous rpc client) we propose three subsequent levels
of changes to the master core architecture.</p>
<p>After making this change we’ll be able to re-evaluate the size of our
thread pool, if we see that we can make most threads in the client
worker pool always idle. In the future we should also investigate making
the rpc client asynchronous as well, so that we can make masterd a lot
smaller in number of threads, and memory size, and thus also easier to
understand, debug, and scale.</p>
<div class="section" id="connection-handling">
<h5>Connection handling<a class="headerlink" href="#connection-handling" title="Permalink to this headline"></a></h5>
<p>We’ll move the main thread of ganeti-masterd to asyncore, so that it can
share the mainloop code with all other Ganeti daemons. Then all luxi
clients will be asyncore clients, and I/O to/from them will be handled
by the master thread asynchronously. Data will be read from the client
sockets as it becomes available, and kept in a buffer, then when a
complete message is found, it’s passed to a client worker thread for
parsing and processing. The client worker thread is responsible for
serializing the reply, which can then be sent asynchronously by the main
thread on the socket.</p>
</div>
<div class="section" id="id1">
<h5>Wait for job change<a class="headerlink" href="#id1" title="Permalink to this headline"></a></h5>
<p>The REQ_WAIT_FOR_JOB_CHANGE luxi request is changed to be
subscription-based, so that the executing thread doesn’t have to be
waiting for the changes to arrive. Threads producing messages (job queue
executors) will make sure that when there is a change another thread is
awakened and delivers it to the waiting clients. This can be either a
dedicated “wait for job changes” thread or pool, or one of the client
workers, depending on what’s easier to implement. In either case the
main asyncore thread will only be involved in pushing of the actual
data, and not in fetching/serializing it.</p>
<p>Other features to look at, when implementing this code are:</p>
<blockquote>
<div><ul class="simple">
<li>Possibility not to need the job lock to know which updates to push:
if the thread producing the data pushed a copy of the update for the
waiting clients, the thread sending it won’t need to acquire the
lock again to fetch the actual data.</li>
<li>Possibility to signal clients about to time out, when no update has
been received, not to despair and to keep waiting (luxi level
keepalive).</li>
<li>Possibility to defer updates if they are too frequent, providing
them at a maximum rate (lower priority).</li>
</ul>
</div></blockquote>
</div>
<div class="section" id="id2">
<h5>Job Queue lock<a class="headerlink" href="#id2" title="Permalink to this headline"></a></h5>
<p>In order to decrease the job queue lock contention, we will change the
code paths in the following ways, initially:</p>
<blockquote>
<div><ul class="simple">
<li>A per-job lock will be introduced. All operations affecting only one
job (for example feedback, starting/finishing notifications,
subscribing to or watching a job) will only require the job lock.
This should be a leaf lock, but if a situation arises in which it
must be acquired together with the global job queue lock the global
one must always be acquired last (for the global section).</li>
<li>The locks will be converted to a sharedlock. Any read-only operation
will be able to proceed in parallel.</li>
<li>During remote update (which happens already per-job) we’ll drop the
job lock level to shared mode, so that activities reading the lock
(for example job change notifications or QueryJobs calls) will be
able to proceed in parallel.</li>
<li>The wait for job changes improvements proposed above will be
implemented.</li>
</ul>
</div></blockquote>
<p>In the future other improvements may include splitting off some of the
work (eg replication of a job to remote nodes) to a separate thread pool
or asynchronous thread, not tied with the code path for answering client
requests or the one executing the “real” work. This can be discussed
again after we used the more granular job queue in production and tested
its benefits.</p>
</div>
</div>
</div>
<div class="section" id="inter-cluster-instance-moves">
<h3><a class="toc-backref" href="#id15">Inter-cluster instance moves</a><a class="headerlink" href="#inter-cluster-instance-moves" title="Permalink to this headline"></a></h3>
<div class="section" id="id3">
<h4><a class="toc-backref" href="#id16">Current state and shortcomings</a><a class="headerlink" href="#id3" title="Permalink to this headline"></a></h4>
<p>With the current design of Ganeti, moving whole instances between
different clusters involves a lot of manual work. There are several ways
to move instances, one of them being to export the instance, manually
copying all data to the new cluster before importing it again. Manual
changes to the instances configuration, such as the IP address, may be
necessary in the new environment. The goal is to improve and automate
this process in Ganeti 2.2.</p>
</div>
<div class="section" id="id4">
<h4><a class="toc-backref" href="#id17">Proposed changes</a><a class="headerlink" href="#id4" title="Permalink to this headline"></a></h4>
<div class="section" id="authorization-authentication-and-security">
<h5>Authorization, Authentication and Security<a class="headerlink" href="#authorization-authentication-and-security" title="Permalink to this headline"></a></h5>
<p>Until now, each Ganeti cluster was a self-contained entity and wouldn’t
talk to other Ganeti clusters. Nodes within clusters only had to trust
the other nodes in the same cluster and the network used for replication
was trusted, too (hence the ability the use a separate, local network
for replication).</p>
<p>For inter-cluster instance transfers this model must be weakened. Nodes
in one cluster will have to talk to nodes in other clusters, sometimes
in other locations and, very important, via untrusted network
connections.</p>
<p>Various option have been considered for securing and authenticating the
data transfer from one machine to another. To reduce the risk of
accidentally overwriting data due to software bugs, authenticating the
arriving data was considered critical. Eventually we decided to use
socat’s OpenSSL options (<code class="docutils literal"><span class="pre">OPENSSL:</span></code>, <code class="docutils literal"><span class="pre">OPENSSL-LISTEN:</span></code> et al), which
provide us with encryption, authentication and authorization when used
with separate keys and certificates.</p>
<p>Combinations of OpenSSH, GnuPG and Netcat were deemed too complex to set
up from within Ganeti. Any solution involving OpenSSH would require a
dedicated user with a home directory and likely automated modifications
to the user’s <code class="docutils literal"><span class="pre">$HOME/.ssh/authorized_keys</span></code> file. When using Netcat,
GnuPG or another encryption method would be necessary to transfer the
data over an untrusted network. socat combines both in one program and
is already a dependency.</p>
<p>Each of the two clusters will have to generate an RSA key. The public
parts are exchanged between the clusters by a third party, such as an
administrator or a system interacting with Ganeti via the remote API
(“third party” from here on). After receiving each other’s public key,
the clusters can start talking to each other.</p>
<p>All encrypted connections must be verified on both sides. Neither side
may accept unverified certificates. The generated certificate should
only be valid for the time necessary to move the instance.</p>
<p>For additional protection of the instance data, the two clusters can
verify the certificates and destination information exchanged via the
third party by checking an HMAC signature using a key shared among the
involved clusters. By default this secret key will be a random string
unique to the cluster, generated by running SHA1 over 20 bytes read from
<code class="docutils literal"><span class="pre">/dev/urandom</span></code> and the administrator must synchronize the secrets
between clusters before instances can be moved. If the third party does
not know the secret, it can’t forge the certificates or redirect the
data. Unless disabled by a new cluster parameter, verifying the HMAC
signatures must be mandatory. The HMAC signature for X509 certificates
will be prepended to the certificate similar to an <span class="target" id="index-0"></span><a class="rfc reference external" href="https://tools.ietf.org/html/rfc822.html"><strong>RFC 822</strong></a> header and
only covers the certificate (from <code class="docutils literal"><span class="pre">-----BEGIN</span> <span class="pre">CERTIFICATE-----</span></code> to
<code class="docutils literal"><span class="pre">-----END</span> <span class="pre">CERTIFICATE-----</span></code>). The header name will be
<code class="docutils literal"><span class="pre">X-Ganeti-Signature</span></code> and its value will have the format
<code class="docutils literal"><span class="pre">$salt/$hash</span></code> (salt and hash separated by slash). The salt may only
contain characters in the range <code class="docutils literal"><span class="pre">[a-zA-Z0-9]</span></code>.</p>
<p>On the web, the destination cluster would be equivalent to an HTTPS
server requiring verifiable client certificates. The browser would be
equivalent to the source cluster and must verify the server’s
certificate while providing a client certificate to the server.</p>
</div>
<div class="section" id="copying-data">
<h5>Copying data<a class="headerlink" href="#copying-data" title="Permalink to this headline"></a></h5>
<p>To simplify the implementation, we decided to operate at a block-device
level only, allowing us to easily support non-DRBD instance moves.</p>
<p>Intra-cluster instance moves will re-use the existing export and import
scripts supplied by instance OS definitions. Unlike simply copying the
raw data, this allows one to use filesystem-specific utilities to dump
only used parts of the disk and to exclude certain disks from the move.
Compression should be used to further reduce the amount of data
transferred.</p>
<p>The export scripts writes all data to stdout and the import script reads
it from stdin again. To avoid copying data and reduce disk space
consumption, everything is read from the disk and sent over the network
directly, where it’ll be written to the new block device directly again.</p>
</div>
<div class="section" id="workflow">
<h5>Workflow<a class="headerlink" href="#workflow" title="Permalink to this headline"></a></h5>
<ol class="arabic simple">
<li>Third party tells source cluster to shut down instance, asks for the
instance specification and for the public part of an encryption key<ul>
<li>Instance information can already be retrieved using an existing API
(<code class="docutils literal"><span class="pre">OpInstanceQueryData</span></code>).</li>
<li>An RSA encryption key and a corresponding self-signed X509
certificate is generated using the “openssl” command. This key will
be used to encrypt the data sent to the destination cluster.<ul>
<li>Private keys never leave the cluster.</li>
<li>The public part (the X509 certificate) is signed using HMAC with
salting and a secret shared between Ganeti clusters.</li>
</ul>
</li>
</ul>
</li>
<li>Third party tells destination cluster to create an instance with the
same specifications as on source cluster and to prepare for an
instance move with the key received from the source cluster and
receives the public part of the destination’s encryption key<ul>
<li>The current API to create instances (<code class="docutils literal"><span class="pre">OpInstanceCreate</span></code>) will be
extended to support an import from a remote cluster.</li>
<li>A valid, unexpired X509 certificate signed with the destination
cluster’s secret will be required. By verifying the signature, we
know the third party didn’t modify the certificate.<ul>
<li>The private keys never leave their cluster, hence the third party
can not decrypt or intercept the instance’s data by modifying the
IP address or port sent by the destination cluster.</li>
</ul>
</li>
<li>The destination cluster generates another key and certificate,
signs and sends it to the third party, who will have to pass it to
the API for exporting an instance (<code class="docutils literal"><span class="pre">OpBackupExport</span></code>). This
certificate is used to ensure we’re sending the disk data to the
correct destination cluster.</li>
<li>Once a disk can be imported, the API sends the destination
information (IP address and TCP port) together with an HMAC
signature to the third party.</li>
</ul>
</li>
<li>Third party hands public part of the destination’s encryption key
together with all necessary information to source cluster and tells
it to start the move<ul>
<li>The existing API for exporting instances (<code class="docutils literal"><span class="pre">OpBackupExport</span></code>)
will be extended to export instances to remote clusters.</li>
</ul>
</li>
<li>Source cluster connects to destination cluster for each disk and
transfers its data using the instance OS definition’s export and
import scripts<ul>
<li>Before starting, the source cluster must verify the HMAC signature
of the certificate and destination information (IP address and TCP
port).</li>
<li>When connecting to the remote machine, strong certificate checks
must be employed.</li>
</ul>
</li>
<li>Due to the asynchronous nature of the whole process, the destination
cluster checks whether all disks have been transferred every time
after transferring a single disk; if so, it destroys the encryption
key</li>
<li>After sending all disks, the source cluster destroys its key</li>
<li>Destination cluster runs OS definition’s rename script to adjust
instance settings if needed (e.g. IP address)</li>
<li>Destination cluster starts the instance if requested at the beginning
by the third party</li>
<li>Source cluster removes the instance if requested</li>
</ol>
</div>
<div class="section" id="instance-move-in-pseudo-code">
<h5>Instance move in pseudo code<a class="headerlink" href="#instance-move-in-pseudo-code" title="Permalink to this headline"></a></h5>
<p>The following pseudo code describes a script moving instances between
clusters and what happens on both clusters.</p>
<ol class="arabic">
<li><p class="first">Script is started, gets the instance name and destination cluster:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="p">(</span><span class="n">instance_name</span><span class="p">,</span> <span class="n">dest_cluster_name</span><span class="p">)</span> <span class="o">=</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="c1"># Get destination cluster object</span>
<span class="n">dest_cluster</span> <span class="o">=</span> <span class="n">db</span><span class="o">.</span><span class="n">FindCluster</span><span class="p">(</span><span class="n">dest_cluster_name</span><span class="p">)</span>

<span class="c1"># Use database to find source cluster</span>
<span class="n">src_cluster</span> <span class="o">=</span> <span class="n">db</span><span class="o">.</span><span class="n">FindClusterByInstance</span><span class="p">(</span><span class="n">instance_name</span><span class="p">)</span>
</pre></div>
</div>
</li>
<li><p class="first">Script tells source cluster to stop instance:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="c1"># Stop instance</span>
<span class="n">src_cluster</span><span class="o">.</span><span class="n">StopInstance</span><span class="p">(</span><span class="n">instance_name</span><span class="p">)</span>

<span class="c1"># Get instance specification (memory, disk, etc.)</span>
<span class="n">inst_spec</span> <span class="o">=</span> <span class="n">src_cluster</span><span class="o">.</span><span class="n">GetInstanceInfo</span><span class="p">(</span><span class="n">instance_name</span><span class="p">)</span>

<span class="p">(</span><span class="n">src_key_name</span><span class="p">,</span> <span class="n">src_cert</span><span class="p">)</span> <span class="o">=</span> <span class="n">src_cluster</span><span class="o">.</span><span class="n">CreateX509Certificate</span><span class="p">()</span>
</pre></div>
</div>
</li>
<li><p class="first"><code class="docutils literal"><span class="pre">CreateX509Certificate</span></code> on source cluster:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">key_file</span> <span class="o">=</span> <span class="n">mkstemp</span><span class="p">()</span>
<span class="n">cert_file</span> <span class="o">=</span> <span class="s2">&quot;</span><span class="si">%s</span><span class="s2">.cert&quot;</span> <span class="o">%</span> <span class="n">key_file</span>
<span class="n">RunCmd</span><span class="p">([</span><span class="s2">&quot;/usr/bin/openssl&quot;</span><span class="p">,</span> <span class="s2">&quot;req&quot;</span><span class="p">,</span> <span class="s2">&quot;-new&quot;</span><span class="p">,</span>
         <span class="s2">&quot;-newkey&quot;</span><span class="p">,</span> <span class="s2">&quot;rsa:1024&quot;</span><span class="p">,</span> <span class="s2">&quot;-days&quot;</span><span class="p">,</span> <span class="s2">&quot;1&quot;</span><span class="p">,</span>
         <span class="s2">&quot;-nodes&quot;</span><span class="p">,</span> <span class="s2">&quot;-x509&quot;</span><span class="p">,</span> <span class="s2">&quot;-batch&quot;</span><span class="p">,</span>
         <span class="s2">&quot;-keyout&quot;</span><span class="p">,</span> <span class="n">key_file</span><span class="p">,</span> <span class="s2">&quot;-out&quot;</span><span class="p">,</span> <span class="n">cert_file</span><span class="p">])</span>

<span class="n">plain_cert</span> <span class="o">=</span> <span class="n">utils</span><span class="o">.</span><span class="n">ReadFile</span><span class="p">(</span><span class="n">cert_file</span><span class="p">)</span>

<span class="c1"># HMAC sign using secret key, this adds a &quot;X-Ganeti-Signature&quot;</span>
<span class="c1"># header to the beginning of the certificate</span>
<span class="n">signed_cert</span> <span class="o">=</span> <span class="n">utils</span><span class="o">.</span><span class="n">SignX509Certificate</span><span class="p">(</span><span class="n">plain_cert</span><span class="p">,</span>
  <span class="n">utils</span><span class="o">.</span><span class="n">ReadFile</span><span class="p">(</span><span class="n">constants</span><span class="o">.</span><span class="n">X509_SIGNKEY_FILE</span><span class="p">))</span>

<span class="c1"># The certificate now looks like the following:</span>
<span class="c1">#</span>
<span class="c1">#   X-Ganeti-Signature: $1234$28676f0516c6ab68062b[…]</span>
<span class="c1">#   -----BEGIN CERTIFICATE-----</span>
<span class="c1">#   MIICsDCCAhmgAwIBAgI[…]</span>
<span class="c1">#   -----END CERTIFICATE-----</span>

<span class="c1"># Return name of key file and signed certificate in PEM format</span>
<span class="k">return</span> <span class="p">(</span><span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">basename</span><span class="p">(</span><span class="n">key_file</span><span class="p">),</span> <span class="n">signed_cert</span><span class="p">)</span>
</pre></div>
</div>
</li>
<li><p class="first">Script creates instance on destination cluster and waits for move to
finish:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">dest_cluster</span><span class="o">.</span><span class="n">CreateInstance</span><span class="p">(</span><span class="n">mode</span><span class="o">=</span><span class="n">constants</span><span class="o">.</span><span class="n">REMOTE_IMPORT</span><span class="p">,</span>
                            <span class="n">spec</span><span class="o">=</span><span class="n">inst_spec</span><span class="p">,</span>
                            <span class="n">source_cert</span><span class="o">=</span><span class="n">src_cert</span><span class="p">)</span>

<span class="c1"># Wait until destination cluster gives us its certificate</span>
<span class="n">dest_cert</span> <span class="o">=</span> <span class="bp">None</span>
<span class="n">disk_info</span> <span class="o">=</span> <span class="p">[]</span>
<span class="k">while</span> <span class="ow">not</span> <span class="p">(</span><span class="n">dest_cert</span> <span class="ow">and</span> <span class="nb">len</span><span class="p">(</span><span class="n">disk_info</span><span class="p">)</span> <span class="o">&lt;</span> <span class="nb">len</span><span class="p">(</span><span class="n">inst_spec</span><span class="o">.</span><span class="n">disks</span><span class="p">)):</span>
  <span class="n">tmp</span> <span class="o">=</span> <span class="n">dest_cluster</span><span class="o">.</span><span class="n">WaitOutput</span><span class="p">()</span>
  <span class="k">if</span> <span class="n">tmp</span> <span class="ow">is</span> <span class="n">Certificate</span><span class="p">:</span>
    <span class="n">dest_cert</span> <span class="o">=</span> <span class="n">tmp</span>
  <span class="k">elif</span> <span class="n">tmp</span> <span class="ow">is</span> <span class="n">DiskInfo</span><span class="p">:</span>
    <span class="c1"># DiskInfo contains destination address and port</span>
    <span class="n">disk_info</span><span class="p">[</span><span class="n">tmp</span><span class="o">.</span><span class="n">index</span><span class="p">]</span> <span class="o">=</span> <span class="n">tmp</span>

<span class="c1"># Tell source cluster to export disks</span>
<span class="k">for</span> <span class="n">disk</span> <span class="ow">in</span> <span class="n">disk_info</span><span class="p">:</span>
  <span class="n">src_cluster</span><span class="o">.</span><span class="n">ExportDisk</span><span class="p">(</span><span class="n">instance_name</span><span class="p">,</span> <span class="n">disk</span><span class="o">=</span><span class="n">disk</span><span class="p">,</span>
                         <span class="n">key_name</span><span class="o">=</span><span class="n">src_key_name</span><span class="p">,</span>
                         <span class="n">dest_cert</span><span class="o">=</span><span class="n">dest_cert</span><span class="p">)</span>

<span class="k">print</span> <span class="p">(</span><span class="s2">&quot;Instance </span><span class="si">%s</span><span class="s2"> sucessfully moved to </span><span class="si">%s</span><span class="s2">&quot;</span> <span class="o">%</span>
       <span class="p">(</span><span class="n">instance_name</span><span class="p">,</span> <span class="n">dest_cluster</span><span class="o">.</span><span class="n">name</span><span class="p">))</span>
</pre></div>
</div>
</li>
<li><p class="first"><code class="docutils literal"><span class="pre">CreateInstance</span></code> on destination cluster:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="c1"># …</span>

<span class="k">if</span> <span class="n">mode</span> <span class="o">==</span> <span class="n">constants</span><span class="o">.</span><span class="n">REMOTE_IMPORT</span><span class="p">:</span>
  <span class="c1"># Make sure certificate was not modified since it was generated by</span>
  <span class="c1"># source cluster (which must use the same secret)</span>
  <span class="k">if</span> <span class="p">(</span><span class="ow">not</span> <span class="n">utils</span><span class="o">.</span><span class="n">VerifySignedX509Cert</span><span class="p">(</span><span class="n">source_cert</span><span class="p">,</span>
        <span class="n">utils</span><span class="o">.</span><span class="n">ReadFile</span><span class="p">(</span><span class="n">constants</span><span class="o">.</span><span class="n">X509_SIGNKEY_FILE</span><span class="p">))):</span>
    <span class="k">raise</span> <span class="n">Error</span><span class="p">(</span><span class="s2">&quot;Certificate not signed with this cluster&#39;s secret&quot;</span><span class="p">)</span>

  <span class="k">if</span> <span class="n">utils</span><span class="o">.</span><span class="n">CheckExpiredX509Cert</span><span class="p">(</span><span class="n">source_cert</span><span class="p">):</span>
    <span class="k">raise</span> <span class="n">Error</span><span class="p">(</span><span class="s2">&quot;X509 certificate is expired&quot;</span><span class="p">)</span>

  <span class="n">source_cert_file</span> <span class="o">=</span> <span class="n">utils</span><span class="o">.</span><span class="n">WriteTempFile</span><span class="p">(</span><span class="n">source_cert</span><span class="p">)</span>

  <span class="c1"># See above for X509 certificate generation and signing</span>
  <span class="p">(</span><span class="n">key_name</span><span class="p">,</span> <span class="n">signed_cert</span><span class="p">)</span> <span class="o">=</span> <span class="n">CreateSignedX509Certificate</span><span class="p">()</span>

  <span class="n">SendToClient</span><span class="p">(</span><span class="s2">&quot;x509-cert&quot;</span><span class="p">,</span> <span class="n">signed_cert</span><span class="p">)</span>

  <span class="k">for</span> <span class="n">disk</span> <span class="ow">in</span> <span class="n">instance</span><span class="o">.</span><span class="n">disks</span><span class="p">:</span>
    <span class="c1"># Start socat</span>
    <span class="n">RunCmd</span><span class="p">((</span><span class="s2">&quot;socat&quot;</span>
            <span class="s2">&quot; OPENSSL-LISTEN:</span><span class="si">%s</span><span class="s2">,…,key=</span><span class="si">%s</span><span class="s2">,cert=</span><span class="si">%s</span><span class="s2">,cafile=</span><span class="si">%s</span><span class="s2">,verify=1&quot;</span>
            <span class="s2">&quot; stdout &gt; /dev/disk…&quot;</span><span class="p">)</span> <span class="o">%</span>
           <span class="n">port</span><span class="p">,</span> <span class="n">GetRsaKeyPath</span><span class="p">(</span><span class="n">key_name</span><span class="p">,</span> <span class="n">private</span><span class="o">=</span><span class="bp">True</span><span class="p">),</span>
           <span class="n">GetRsaKeyPath</span><span class="p">(</span><span class="n">key_name</span><span class="p">,</span> <span class="n">private</span><span class="o">=</span><span class="bp">False</span><span class="p">),</span> <span class="n">src_cert_file</span><span class="p">)</span>
    <span class="n">SendToClient</span><span class="p">(</span><span class="s2">&quot;send-disk-to&quot;</span><span class="p">,</span> <span class="n">disk</span><span class="p">,</span> <span class="n">ip_address</span><span class="p">,</span> <span class="n">port</span><span class="p">)</span>

  <span class="n">DestroyX509Cert</span><span class="p">(</span><span class="n">key_name</span><span class="p">)</span>

  <span class="n">RunRenameScript</span><span class="p">(</span><span class="n">instance_name</span><span class="p">)</span>
</pre></div>
</div>
</li>
<li><p class="first"><code class="docutils literal"><span class="pre">ExportDisk</span></code> on source cluster:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="c1"># Make sure certificate was not modified since it was generated by</span>
<span class="c1"># destination cluster (which must use the same secret)</span>
<span class="k">if</span> <span class="p">(</span><span class="ow">not</span> <span class="n">utils</span><span class="o">.</span><span class="n">VerifySignedX509Cert</span><span class="p">(</span><span class="n">cert_pem</span><span class="p">,</span>
      <span class="n">utils</span><span class="o">.</span><span class="n">ReadFile</span><span class="p">(</span><span class="n">constants</span><span class="o">.</span><span class="n">X509_SIGNKEY_FILE</span><span class="p">))):</span>
  <span class="k">raise</span> <span class="n">Error</span><span class="p">(</span><span class="s2">&quot;Certificate not signed with this cluster&#39;s secret&quot;</span><span class="p">)</span>

<span class="k">if</span> <span class="n">utils</span><span class="o">.</span><span class="n">CheckExpiredX509Cert</span><span class="p">(</span><span class="n">cert_pem</span><span class="p">):</span>
  <span class="k">raise</span> <span class="n">Error</span><span class="p">(</span><span class="s2">&quot;X509 certificate is expired&quot;</span><span class="p">)</span>

<span class="n">dest_cert_file</span> <span class="o">=</span> <span class="n">utils</span><span class="o">.</span><span class="n">WriteTempFile</span><span class="p">(</span><span class="n">cert_pem</span><span class="p">)</span>

<span class="c1"># Start socat</span>
<span class="n">RunCmd</span><span class="p">((</span><span class="s2">&quot;socat stdin&quot;</span>
        <span class="s2">&quot; OPENSSL:</span><span class="si">%s</span><span class="s2">:</span><span class="si">%s</span><span class="s2">,…,key=</span><span class="si">%s</span><span class="s2">,cert=</span><span class="si">%s</span><span class="s2">,cafile=</span><span class="si">%s</span><span class="s2">,verify=1&quot;</span>
        <span class="s2">&quot; &lt; /dev/disk…&quot;</span><span class="p">)</span> <span class="o">%</span>
       <span class="n">disk</span><span class="o">.</span><span class="n">host</span><span class="p">,</span> <span class="n">disk</span><span class="o">.</span><span class="n">port</span><span class="p">,</span>
       <span class="n">GetRsaKeyPath</span><span class="p">(</span><span class="n">key_name</span><span class="p">,</span> <span class="n">private</span><span class="o">=</span><span class="bp">True</span><span class="p">),</span>
       <span class="n">GetRsaKeyPath</span><span class="p">(</span><span class="n">key_name</span><span class="p">,</span> <span class="n">private</span><span class="o">=</span><span class="bp">False</span><span class="p">),</span> <span class="n">dest_cert_file</span><span class="p">)</span>

<span class="k">if</span> <span class="n">instance</span><span class="o">.</span><span class="n">all_disks_done</span><span class="p">:</span>
  <span class="n">DestroyX509Cert</span><span class="p">(</span><span class="n">key_name</span><span class="p">)</span>
</pre></div>
</div>
</li>
</ol>
</div>
<div class="section" id="miscellaneous-notes">
<h5>Miscellaneous notes<a class="headerlink" href="#miscellaneous-notes" title="Permalink to this headline"></a></h5>
<ul class="simple">
<li>A very similar system could also be used for instance exports within
the same cluster. Currently OpenSSH is being used, but could be
replaced by socat and SSL/TLS.</li>
<li>During the design of intra-cluster instance moves we also discussed
encrypting instance exports using GnuPG.</li>
<li>While most instances should have exactly the same configuration as
on the source cluster, setting them up with a different disk layout
might be helpful in some use-cases.</li>
<li>A cleanup operation, similar to the one available for failed instance
migrations, should be provided.</li>
<li><code class="docutils literal"><span class="pre">ganeti-watcher</span></code> should remove instances pending a move from another
cluster after a certain amount of time. This takes care of failures
somewhere in the process.</li>
<li>RSA keys can be generated using the existing
<code class="docutils literal"><span class="pre">bootstrap.GenerateSelfSignedSslCert</span></code> function, though it might be
useful to not write both parts into a single file, requiring small
changes to the function. The public part always starts with
<code class="docutils literal"><span class="pre">-----BEGIN</span> <span class="pre">CERTIFICATE-----</span></code> and ends with <code class="docutils literal"><span class="pre">-----END</span>
<span class="pre">CERTIFICATE-----</span></code>.</li>
<li>The source and destination cluster might be different when it comes
to available hypervisors, kernels, etc. The destination cluster should
refuse to accept an instance move if it can’t fulfill an instance’s
requirements.</li>
</ul>
</div>
</div>
</div>
<div class="section" id="privilege-separation">
<h3><a class="toc-backref" href="#id18">Privilege separation</a><a class="headerlink" href="#privilege-separation" title="Permalink to this headline"></a></h3>
<div class="section" id="id5">
<h4><a class="toc-backref" href="#id19">Current state and shortcomings</a><a class="headerlink" href="#id5" title="Permalink to this headline"></a></h4>
<p>All Ganeti daemons are run under the user root. This is not ideal from a
security perspective as for possible exploitation of any daemon the user
has full access to the system.</p>
<p>In order to overcome this situation we’ll allow Ganeti to run its daemon
under different users and a dedicated group. This also will allow some
side effects, like letting the user run some <code class="docutils literal"><span class="pre">gnt-*</span></code> commands if one
is in the same group.</p>
</div>
<div class="section" id="implementation">
<h4><a class="toc-backref" href="#id20">Implementation</a><a class="headerlink" href="#implementation" title="Permalink to this headline"></a></h4>
<p>For Ganeti 2.2 the implementation will be focused on a the RAPI daemon
only. This involves changes to <code class="docutils literal"><span class="pre">daemons.py</span></code> so it’s possible to drop
privileges on daemonize the process. Though, this will be a short term
solution which will be replaced by a privilege drop already on daemon
startup in Ganeti 2.3.</p>
<p>It also needs changes in the master daemon to create the socket with new
permissions/owners to allow RAPI access. There will be no other
permission/owner changes in the file structure as the RAPI daemon is
started with root permission. In that time it will read all needed files
and then drop privileges before contacting the master daemon.</p>
</div>
</div>
</div>
<div class="section" id="feature-changes">
<h2><a class="toc-backref" href="#id21">Feature changes</a><a class="headerlink" href="#feature-changes" title="Permalink to this headline"></a></h2>
<div class="section" id="kvm-security">
<h3><a class="toc-backref" href="#id22">KVM Security</a><a class="headerlink" href="#kvm-security" title="Permalink to this headline"></a></h3>
<div class="section" id="id6">
<h4><a class="toc-backref" href="#id23">Current state and shortcomings</a><a class="headerlink" href="#id6" title="Permalink to this headline"></a></h4>
<p>Currently all kvm processes run as root. Taking ownership of the
hypervisor process, from inside a virtual machine, would mean a full
compromise of the whole Ganeti cluster, knowledge of all Ganeti
authentication secrets, full access to all running instances, and the
option of subverting other basic services on the cluster (eg: ssh).</p>
</div>
<div class="section" id="id7">
<h4><a class="toc-backref" href="#id24">Proposed changes</a><a class="headerlink" href="#id7" title="Permalink to this headline"></a></h4>
<p>We would like to decrease the surface of attack available if an
hypervisor is compromised. We can do so adding different features to
Ganeti, which will allow restricting the broken hypervisor
possibilities, in the absence of a local privilege escalation attack, to
subvert the node.</p>
<div class="section" id="dropping-privileges-in-kvm-to-a-single-user-easy">
<h5>Dropping privileges in kvm to a single user (easy)<a class="headerlink" href="#dropping-privileges-in-kvm-to-a-single-user-easy" title="Permalink to this headline"></a></h5>
<p>By passing the <code class="docutils literal"><span class="pre">-runas</span></code> option to kvm, we can make it drop privileges.
The user can be chosen by an hypervisor parameter, so that each instance
can have its own user, but by default they will all run under the same
one. It should be very easy to implement, and can easily be backported
to 2.1.X.</p>
<p>This mode protects the Ganeti cluster from a subverted hypervisor, but
doesn’t protect the instances between each other, unless care is taken
to specify a different user for each. This would prevent the worst
attacks, including:</p>
<ul class="simple">
<li>logging in to other nodes</li>
<li>administering the Ganeti cluster</li>
<li>subverting other services</li>
</ul>
<p>But the following would remain an option:</p>
<ul class="simple">
<li>terminate other VMs (but not start them again, as that requires root
privileges to set up networking) (unless different users are used)</li>
<li>trace other VMs, and probably subvert them and access their data
(unless different users are used)</li>
<li>send network traffic from the node</li>
<li>read unprotected data on the node filesystem</li>
</ul>
</div>
<div class="section" id="running-kvm-in-a-chroot-slightly-harder">
<h5>Running kvm in a chroot (slightly harder)<a class="headerlink" href="#running-kvm-in-a-chroot-slightly-harder" title="Permalink to this headline"></a></h5>
<p>By passing the <code class="docutils literal"><span class="pre">-chroot</span></code> option to kvm, we can restrict the kvm
process in its own (possibly empty) root directory. We need to set this
area up so that the instance disks and control sockets are accessible,
so it would require slightly more work at the Ganeti level.</p>
<p>Breaking out in a chroot would mean:</p>
<ul class="simple">
<li>a lot less options to find a local privilege escalation vector</li>
<li>the impossibility to write local data, if the chroot is set up
correctly</li>
<li>the impossibility to read filesystem data on the host</li>
</ul>
<p>It would still be possible though to:</p>
<ul class="simple">
<li>terminate other VMs</li>
<li>trace other VMs, and possibly subvert them (if a tracer can be
installed in the chroot)</li>
<li>send network traffic from the node</li>
</ul>
</div>
<div class="section" id="running-kvm-with-a-pool-of-users-slightly-harder">
<h5>Running kvm with a pool of users (slightly harder)<a class="headerlink" href="#running-kvm-with-a-pool-of-users-slightly-harder" title="Permalink to this headline"></a></h5>
<p>If rather than passing a single user as an hypervisor parameter, we have
a pool of useable ones, we can dynamically choose a free one to use and
thus guarantee that each machine will be separate from the others,
without putting the burden of this on the cluster administrator.</p>
<p>This would mean interfering between machines would be impossible, and
can still be combined with the chroot benefits.</p>
</div>
<div class="section" id="running-iptables-rules-to-limit-network-interaction-easy">
<h5>Running iptables rules to limit network interaction (easy)<a class="headerlink" href="#running-iptables-rules-to-limit-network-interaction-easy" title="Permalink to this headline"></a></h5>
<p>These don’t need to be handled by Ganeti, but we can ship examples. If
the users used to run VMs would be blocked from sending some or all
network traffic, it would become impossible for a broken into hypervisor
to send arbitrary data on the node network, which is especially useful
when the instance and the node network are separated (using ganeti-nbma
or a separate set of network interfaces), or when a separate replication
network is maintained. We need to experiment to see how much restriction
we can properly apply, without limiting the instance legitimate traffic.</p>
</div>
<div class="section" id="running-kvm-inside-a-container-even-harder">
<h5>Running kvm inside a container (even harder)<a class="headerlink" href="#running-kvm-inside-a-container-even-harder" title="Permalink to this headline"></a></h5>
<p>Recent linux kernels support different process namespaces through
control groups. PIDs, users, filesystems and even network interfaces can
be separated. If we can set up ganeti to run kvm in a separate container
we could insulate all the host process from being even visible if the
hypervisor gets broken into. Most probably separating the network
namespace would require one extra hop in the host, through a veth
interface, thus reducing performance, so we may want to avoid that, and
just rely on iptables.</p>
</div>
</div>
<div class="section" id="implementation-plan">
<h4><a class="toc-backref" href="#id25">Implementation plan</a><a class="headerlink" href="#implementation-plan" title="Permalink to this headline"></a></h4>
<p>We will first implement dropping privileges for kvm processes as a
single user, and most probably backport it to 2.1. Then we’ll ship
example iptables rules to show how the user can be limited in its
network activities.  After that we’ll implement chroot restriction for
kvm processes, and extend the user limitation to use a user pool.</p>
<p>Finally we’ll look into namespaces and containers, although that might
slip after the 2.2 release.</p>
</div>
</div>
<div class="section" id="new-os-states">
<h3><a class="toc-backref" href="#id26">New OS states</a><a class="headerlink" href="#new-os-states" title="Permalink to this headline"></a></h3>
<p>Separate from the OS external changes, described below, we’ll add some
internal changes to the OS.</p>
<div class="section" id="id8">
<h4><a class="toc-backref" href="#id27">Current state and shortcomings</a><a class="headerlink" href="#id8" title="Permalink to this headline"></a></h4>
<p>There are two issues related to the handling of the OSes.</p>
<p>First, it’s impossible to disable an OS for new instances, since that
will also break reinstallations and renames of existing instances. To
phase out an OS definition, without actually having to modify the OS
scripts, it would be ideal to be able to restrict new installations but
keep the rest of the functionality available.</p>
<p>Second, <code class="docutils literal"><span class="pre">gnt-instance</span> <span class="pre">reinstall</span> <span class="pre">--select-os</span></code> shows all the OSes
available on the clusters. Some OSes might exist only for debugging and
diagnose, and not for end-user availability. For this, it would be
useful to “hide” a set of OSes, but keep it otherwise functional.</p>
</div>
<div class="section" id="id9">
<h4><a class="toc-backref" href="#id28">Proposed changes</a><a class="headerlink" href="#id9" title="Permalink to this headline"></a></h4>
<p>Two new cluster-level attributes will be added, holding the list of OSes
hidden from the user and respectively the list of OSes which are
blacklisted from new installations.</p>
<p>These lists will be modifiable via <code class="docutils literal"><span class="pre">gnt-os</span> <span class="pre">modify</span></code> (implemented via
<code class="docutils literal"><span class="pre">OpClusterSetParams</span></code>), such that even not-yet-existing OSes can be
preseeded into a given state.</p>
<p>For the hidden OSes, they are fully functional except that they are not
returned in the default OS list (as computed via <code class="docutils literal"><span class="pre">OpOsDiagnose</span></code>),
unless the hidden state is requested.</p>
<p>For the blacklisted OSes, they are also not shown (unless the
blacklisted state is requested), and they are also prevented from
installation via <code class="docutils literal"><span class="pre">OpInstanceCreate</span></code> (in create mode).</p>
<p>Both these attributes are per-OS, not per-variant. Thus they apply to
all of an OS’ variants, and it’s impossible to blacklist or hide just
one variant. Further improvements might allow a given OS variant to be
blacklisted, as opposed to whole OSes.</p>
</div>
</div>
</div>
<div class="section" id="external-interface-changes">
<h2><a class="toc-backref" href="#id29">External interface changes</a><a class="headerlink" href="#external-interface-changes" title="Permalink to this headline"></a></h2>
<div class="section" id="os-api">
<h3><a class="toc-backref" href="#id30">OS API</a><a class="headerlink" href="#os-api" title="Permalink to this headline"></a></h3>
<p>The OS variants implementation in Ganeti 2.1 didn’t prove to be useful
enough to alleviate the need to hack around the Ganeti API in order to
provide flexible OS parameters.</p>
<p>As such, for Ganeti 2.2 we will provide support for arbitrary OS
parameters. However, since OSes are not registered in Ganeti, but
instead discovered at runtime, the interface is not entirely
straightforward.</p>
<p>Furthermore, to support the system administrator in keeping OSes
properly in sync across the nodes of a cluster, Ganeti will also verify
(if existing) the consistence of a new <code class="docutils literal"><span class="pre">os_version</span></code> file.</p>
<p>These changes to the OS API will bump the API version to 20.</p>
<div class="section" id="os-version">
<h4><a class="toc-backref" href="#id31">OS version</a><a class="headerlink" href="#os-version" title="Permalink to this headline"></a></h4>
<p>A new <code class="docutils literal"><span class="pre">os_version</span></code> file will be supported by Ganeti. This file is not
required, but if existing, its contents will be checked for consistency
across nodes. The file should hold only one line of text (any extra data
will be discarded), and its contents will be shown in the OS information
and diagnose commands.</p>
<p>It is recommended that OS authors increase the contents of this file for
any changes; at a minimum, modifications that change the behaviour of
import/export scripts must increase the version, since they break
intra-cluster migration.</p>
</div>
<div class="section" id="parameters">
<h4><a class="toc-backref" href="#id32">Parameters</a><a class="headerlink" href="#parameters" title="Permalink to this headline"></a></h4>
<p>The interface between Ganeti and the OS scripts will be based on
environment variables, and as such the parameters and their values will
need to be valid in this context.</p>
<div class="section" id="names">
<h5>Names<a class="headerlink" href="#names" title="Permalink to this headline"></a></h5>
<p>The parameter names will be declared in a new file, <code class="docutils literal"><span class="pre">parameters.list</span></code>,
together with a one-line documentation (whitespace-separated). Example:</p>
<div class="highlight-text"><div class="highlight"><pre><span></span>$ cat parameters.list
ns1    Specifies the first name server to add to /etc/resolv.conf
extra_packages  Specifies additional packages to install
rootfs_size     Specifies the root filesystem size (the rest will be left unallocated)
track  Specifies the distribution track, one of &#39;stable&#39;, &#39;testing&#39; or &#39;unstable&#39;
</pre></div>
</div>
<p>As seen above, the documentation can be separate via multiple
spaces/tabs from the names.</p>
<p>The parameter names as read from the file will be used for the command
line interface in lowercased form; as such, there shouldn’t be any two
parameters which differ in case only.</p>
</div>
<div class="section" id="values">
<h5>Values<a class="headerlink" href="#values" title="Permalink to this headline"></a></h5>
<p>The values of the parameters are, from Ganeti’s point of view,
completely freeform. If a given parameter has, from the OS’ point of
view, a fixed set of valid values, these should be documented as such
and verified by the OS, but Ganeti will not handle such parameters
specially.</p>
<p>An empty value must be handled identically as a missing parameter. In
other words, the validation script should only test for non-empty
values, and not for declared versus undeclared parameters.</p>
<p>Furthermore, each parameter should have an (internal to the OS) default
value, that will be used if not passed from Ganeti. More precisely, it
should be possible for any parameter to specify a value that will have
the same effect as not passing the parameter, and no in no case should
the absence of a parameter be treated as an exceptional case (outside
the value space).</p>
<div class="section" id="environment-variables">
<h6>Environment variables<a class="headerlink" href="#environment-variables" title="Permalink to this headline"></a></h6>
<p>The parameters will be exposed in the environment upper-case and
prefixed with the string <code class="docutils literal"><span class="pre">OSP_</span></code>. For example, a parameter declared in
the ‘parameters’ file as <code class="docutils literal"><span class="pre">ns1</span></code> will appear in the environment as the
variable <code class="docutils literal"><span class="pre">OSP_NS1</span></code>.</p>
</div>
</div>
<div class="section" id="validation">
<h5>Validation<a class="headerlink" href="#validation" title="Permalink to this headline"></a></h5>
<p>For the purpose of parameter name/value validation, the OS scripts
<em>must</em> provide an additional script, named <code class="docutils literal"><span class="pre">verify</span></code>. This script will
be called with the argument <code class="docutils literal"><span class="pre">parameters</span></code>, and all the parameters will
be passed in via environment variables, as described above.</p>
<p>The script should signify result/failure based on its exit code, and
show explanatory messages either on its standard output or standard
error. These messages will be passed on to the master, and stored as in
the OpCode result/error message.</p>
<p>The parameters must be constructed to be independent of the instance
specifications. In general, the validation script will only be called
with the parameter variables set, but not with the normal per-instance
variables, in order for Ganeti to be able to validate default parameters
too, when they change. Validation will only be performed on one cluster
node, and it will be up to the ganeti administrator to keep the OS
scripts in sync between all nodes.</p>
</div>
<div class="section" id="instance-operations">
<h5>Instance operations<a class="headerlink" href="#instance-operations" title="Permalink to this headline"></a></h5>
<p>The parameters will be passed, as described above, to all the other
instance operations (creation, import, export). Ideally, these scripts
will not abort with parameter validation errors, if the <code class="docutils literal"><span class="pre">verify</span></code>
script has verified them correctly.</p>
<p>Note: when changing an instance’s OS type, any OS parameters defined at
instance level will be kept as-is. If the parameters differ between the
new and the old OS, the user should manually remove/update them as
needed.</p>
</div>
<div class="section" id="declaration-and-modification">
<h5>Declaration and modification<a class="headerlink" href="#declaration-and-modification" title="Permalink to this headline"></a></h5>
<p>Since the OSes are not registered in Ganeti, we will only make a ‘weak’
link between the parameters as declared in Ganeti and the actual OSes
existing on the cluster.</p>
<p>It will be possible to declare parameters either globally, per cluster
(where they are indexed per OS/variant), or individually, per
instance. The declaration of parameters will not be tied to current
existing OSes. When specifying a parameter, if the OS exists, it will be
validated; if not, then it will simply be stored as-is.</p>
<p>A special note is that it will not be possible to ‘unset’ at instance
level a parameter that is declared globally. Instead, at instance level
the parameter should be given an explicit value, or the default value as
explained above.</p>
</div>
<div class="section" id="cli-interface">
<h5>CLI interface<a class="headerlink" href="#cli-interface" title="Permalink to this headline"></a></h5>
<p>The modification of global (default) parameters will be done via the
<code class="docutils literal"><span class="pre">gnt-os</span></code> command, and the per-instance parameters via the
<code class="docutils literal"><span class="pre">gnt-instance</span></code> command. Both these commands will take an addition
<code class="docutils literal"><span class="pre">--os-parameters</span></code> or <code class="docutils literal"><span class="pre">-O</span></code> flag that specifies the parameters in the
familiar comma-separated, key=value format. For removing a parameter, a
<code class="docutils literal"><span class="pre">-key</span></code> syntax will be used, e.g.:</p>
<div class="highlight-text"><div class="highlight"><pre><span></span># initial modification
$ gnt-instance modify -O use_dchp=true instance1
# later revert (to the cluster default, or the OS default if not
# defined at cluster level)
$ gnt-instance modify -O -use_dhcp instance1
</pre></div>
</div>
</div>
<div class="section" id="internal-storage">
<h5>Internal storage<a class="headerlink" href="#internal-storage" title="Permalink to this headline"></a></h5>
<p>Internally, the OS parameters will be stored in a new <code class="docutils literal"><span class="pre">osparams</span></code>
attribute. The global parameters will be stored on the cluster object,
and the value of this attribute will be a dictionary indexed by OS name
(this also accepts an OS+variant name, which will override a simple OS
name, see below), and for values the key/name dictionary. For the
instances, the value will be directly the key/name dictionary.</p>
</div>
<div class="section" id="overriding-rules">
<h5>Overriding rules<a class="headerlink" href="#overriding-rules" title="Permalink to this headline"></a></h5>
<p>Any instance-specific parameters will override any variant-specific
parameters, which in turn will override any global parameters. The
global parameters, in turn, override the built-in defaults (of the OS
scripts).</p>
</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 2.2 design</a><ul>
<li><a class="reference internal" href="#core-changes">Core changes</a><ul>
<li><a class="reference internal" href="#master-daemon-scaling-improvements">Master Daemon Scaling improvements</a><ul>
<li><a class="reference internal" href="#current-state-and-shortcomings">Current state and shortcomings</a><ul>
<li><a class="reference internal" href="#core-daemon-connection-handling">Core daemon connection handling</a></li>
<li><a class="reference internal" href="#wait-for-job-change">Wait for job change</a></li>
<li><a class="reference internal" href="#job-queue-lock">Job Queue lock</a></li>
</ul>
</li>
<li><a class="reference internal" href="#proposed-changes">Proposed changes</a><ul>
<li><a class="reference internal" href="#connection-handling">Connection handling</a></li>
<li><a class="reference internal" href="#id1">Wait for job change</a></li>
<li><a class="reference internal" href="#id2">Job Queue lock</a></li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#inter-cluster-instance-moves">Inter-cluster instance moves</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="#authorization-authentication-and-security">Authorization, Authentication and Security</a></li>
<li><a class="reference internal" href="#copying-data">Copying data</a></li>
<li><a class="reference internal" href="#workflow">Workflow</a></li>
<li><a class="reference internal" href="#instance-move-in-pseudo-code">Instance move in pseudo code</a></li>
<li><a class="reference internal" href="#miscellaneous-notes">Miscellaneous notes</a></li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#privilege-separation">Privilege separation</a><ul>
<li><a class="reference internal" href="#id5">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#implementation">Implementation</a></li>
</ul>
</li>
</ul>
</li>
<li><a class="reference internal" href="#feature-changes">Feature changes</a><ul>
<li><a class="reference internal" href="#kvm-security">KVM Security</a><ul>
<li><a class="reference internal" href="#id6">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#id7">Proposed changes</a><ul>
<li><a class="reference internal" href="#dropping-privileges-in-kvm-to-a-single-user-easy">Dropping privileges in kvm to a single user (easy)</a></li>
<li><a class="reference internal" href="#running-kvm-in-a-chroot-slightly-harder">Running kvm in a chroot (slightly harder)</a></li>
<li><a class="reference internal" href="#running-kvm-with-a-pool-of-users-slightly-harder">Running kvm with a pool of users (slightly harder)</a></li>
<li><a class="reference internal" href="#running-iptables-rules-to-limit-network-interaction-easy">Running iptables rules to limit network interaction (easy)</a></li>
<li><a class="reference internal" href="#running-kvm-inside-a-container-even-harder">Running kvm inside a container (even harder)</a></li>
</ul>
</li>
<li><a class="reference internal" href="#implementation-plan">Implementation plan</a></li>
</ul>
</li>
<li><a class="reference internal" href="#new-os-states">New OS states</a><ul>
<li><a class="reference internal" href="#id8">Current state and shortcomings</a></li>
<li><a class="reference internal" href="#id9">Proposed changes</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="#os-version">OS version</a></li>
<li><a class="reference internal" href="#parameters">Parameters</a><ul>
<li><a class="reference internal" href="#names">Names</a></li>
<li><a class="reference internal" href="#values">Values</a><ul>
<li><a class="reference internal" href="#environment-variables">Environment variables</a></li>
</ul>
</li>
<li><a class="reference internal" href="#validation">Validation</a></li>
<li><a class="reference internal" href="#instance-operations">Instance operations</a></li>
<li><a class="reference internal" href="#declaration-and-modification">Declaration and modification</a></li>
<li><a class="reference internal" href="#cli-interface">CLI interface</a></li>
<li><a class="reference internal" href="#internal-storage">Internal storage</a></li>
<li><a class="reference internal" href="#overriding-rules">Overriding rules</a></li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
</ul>

  <h4>Previous topic</h4>
  <p class="topless"><a href="design-2.1.html"
                        title="previous chapter">Ganeti 2.1 design</a></p>
  <h4>Next topic</h4>
  <p class="topless"><a href="design-2.3.html"
                        title="next chapter">Ganeti 2.3 design</a></p>
  <div role="note" aria-label="source link">
    <h3>This Page</h3>
    <ul class="this-page-menu">
      <li><a href="_sources/design-2.2.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-2.3.html" title="Ganeti 2.3 design"
             >next</a></li>
        <li class="right" >
          <a href="design-2.1.html" title="Ganeti 2.1 design"
             >previous</a> |</li>
        <li class="nav-item nav-item-0"><a href="index.html">Ganeti 2.16.0~rc2 documentation</a> &#187;</li> 
      </ul>
    </div>
    <div class="footer" role="contentinfo">
        &#169; 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>