/usr/share/doc/slony1-2-doc/adminguide/failover.html is in slony1-2-doc 2.2.5-2.
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 | <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN""http://www.w3.org/TR/html4/loose.dtd">
<HTML
><HEAD
><TITLE
>Doing switchover and failover with Slony-I</TITLE
><META
NAME="GENERATOR"
CONTENT="Modular DocBook HTML Stylesheet Version 1.79"><LINK
REV="MADE"
HREF="mailto:slony1-general@lists.slony.info"><LINK
REL="HOME"
TITLE="Slony-I 2.2.5 Documentation"
HREF="index.html"><LINK
REL="UP"
TITLE="Administration Tasks"
HREF="administration.html"><LINK
REL="PREVIOUS"
TITLE="Database Schema Changes (DDL)"
HREF="ddlchanges.html"><LINK
REL="NEXT"
TITLE="Advanced Concepts"
HREF="advanced.html"><LINK
REL="STYLESHEET"
TYPE="text/css"
HREF="stylesheet.css"><META
HTTP-EQUIV="Content-Type"
CONTENT="text/html; charset=ISO-8859-1"></HEAD
><BODY
CLASS="SECT1"
><DIV
CLASS="NAVHEADER"
><TABLE
SUMMARY="Header navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TH
COLSPAN="5"
ALIGN="center"
VALIGN="bottom"
><SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> 2.2.5 Documentation</TH
></TR
><TR
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="top"
><A
HREF="ddlchanges.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="10%"
ALIGN="left"
VALIGN="top"
><A
HREF="administration.html"
>Fast Backward</A
></TD
><TD
WIDTH="60%"
ALIGN="center"
VALIGN="bottom"
>Chapter 3. Administration Tasks</TD
><TD
WIDTH="10%"
ALIGN="right"
VALIGN="top"
><A
HREF="administration.html"
>Fast Forward</A
></TD
><TD
WIDTH="10%"
ALIGN="right"
VALIGN="top"
><A
HREF="advanced.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
></TABLE
><HR
ALIGN="LEFT"
WIDTH="100%"></DIV
><DIV
CLASS="SECT1"
><H1
CLASS="SECT1"
><A
NAME="FAILOVER"
>3.4. Doing switchover and failover with <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
></A
></H1
><A
NAME="AEN1223"
></A
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN1226"
>3.4.1. Foreword</A
></H2
><P
><SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> is an asynchronous replication system. Because of
that, it is almost certain that at the moment the current origin of a
set fails, the final transactions committed at the origin will have
not yet propagated to the subscribers. Systems are particularly
likely to fail under heavy load; that is one of the corollaries of
Murphy's Law. Therefore the principal goal is to
<SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>prevent</I
></SPAN
> the main server from failing. The best
way to do that is frequent maintenance.</P
><P
> Opening the case of a running server is not exactly what we
should consider a <SPAN
CLASS="QUOTE"
>"professional"</SPAN
> way to do system
maintenance. And interestingly, those users who found it valuable to
use replication for backup and failover purposes are the very ones
that have the lowest tolerance for terms like <SPAN
CLASS="QUOTE"
>"system
downtime."</SPAN
> To help support these requirements, <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> not
only offers failover capabilities, but also the notion of controlled
origin transfer.</P
><P
> It is assumed in this document that the reader is familiar with
the <A
HREF="slonik.html"
><SPAN
CLASS="APPLICATION"
>slonik</SPAN
></A
> utility and knows at least how to set up a
simple 2 node replication system with <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
>.</P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN1238"
>3.4.2. Controlled Switchover</A
></H2
><A
NAME="AEN1240"
></A
><P
> We assume a current <SPAN
CLASS="QUOTE"
>"origin"</SPAN
> as node1 with one
<SPAN
CLASS="QUOTE"
>"subscriber"</SPAN
> as node2 (<SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>e.g.</I
></SPAN
> -
slave). A web application on a third server is accessing the database
on node1. Both databases are up and running and replication is more
or less in sync. We do controlled switchover using <A
HREF="stmtmoveset.html"
>SLONIK MOVE SET</A
>.</P
><P
></P
><UL
><LI
><P
> At the time of this writing switchover to another
server requires the application to reconnect to the new database. So
in order to avoid any complications, we simply shut down the web
server. Users who use <SPAN
CLASS="APPLICATION"
>pg_pool</SPAN
> for the
applications database connections merely have to shut down the
pool.</P
><P
> What needs to be done, here, is highly dependent on the way
that the application(s) that use the database are configured. The
general point is thus: Applications that were connected to the old
database must drop those connections and establish new connections to
the database that has been promoted to the <SPAN
CLASS="QUOTE"
>"master"</SPAN
> role. There
are a number of ways that this may be configured, and therefore, a
number of possible methods for accomplishing the change:</P
><P
></P
><UL
><LI
><P
> The application may store the name of the database in
a file.</P
><P
> In that case, the reconfiguration may require changing the
value in the file, and stopping and restarting the application to get
it to point to the new location.</P
></LI
><LI
><P
> A clever usage of DNS might involve creating a CNAME
<A
HREF="http://www.iana.org/assignments/dns-parameters"
TARGET="_top"
> DNS
record </A
> that establishes a name for the application to use to
reference the node that is in the <SPAN
CLASS="QUOTE"
>"master"</SPAN
> role.</P
><P
> In that case, reconfiguration would require changing the CNAME
to point to the new server, and possibly restarting the application to
refresh database connections.</P
></LI
><LI
><P
> If you are using <SPAN
CLASS="APPLICATION"
>pg_pool</SPAN
> or some
similar <SPAN
CLASS="QUOTE"
>"connection pool manager,"</SPAN
> then the reconfiguration
involves reconfiguring this management tool, but is otherwise similar
to the DNS/CNAME example above. </P
></LI
></UL
><P
> Whether or not the application that accesses the database needs
to be restarted depends on how it is coded to cope with failed
database connections; if, after encountering an error it tries
re-opening them, then there may be no need to restart it. </P
></LI
><LI
><P
> A small <A
HREF="slonik.html"
><SPAN
CLASS="APPLICATION"
>slonik</SPAN
></A
> script executes the
following commands:
</P><PRE
CLASS="PROGRAMLISTING"
>lock set (id = 1, origin = 1);
wait for event (origin = 1, confirmed = 2, wait on=1);
move set (id = 1, old origin = 1, new origin = 2);
wait for event (origin = 1, confirmed = 2, wait on=1);</PRE
><P></P
><P
> After these commands, the origin (master role) of data set 1
has been transferred to node2. And it is not simply transferred; it
is done in a fashion such that node1 becomes a fully synchronized
subscriber, actively replicating the set. So the two nodes have
switched roles completely.</P
></LI
><LI
><P
> After reconfiguring the web application (or
<SPAN
CLASS="APPLICATION"
><A
HREF="help.html#PGPOOL"
> pgpool </A
></SPAN
>) to
connect to the database on node2, the web server is restarted and
resumes normal operation.</P
><P
> Done in one shell script, that does the application shutdown,
<SPAN
CLASS="APPLICATION"
>slonik</SPAN
>, move config files and startup all
together, this entire procedure is likely to take less than 10
seconds.</P
></LI
></UL
><P
> You may now simply shutdown the server hosting node1 and do
whatever is required to maintain the server. When <A
HREF="slon.html"
><SPAN
CLASS="APPLICATION"
>slon</SPAN
></A
> node1 is restarted later, it will start replicating
again, and soon catch up. At this point the procedure to switch
origins is executed again to restore the original
configuration.</P
><P
> This is the preferred way to handle things; it runs quickly,
under control of the administrators, and there is no need for there to
be any loss of data.</P
><P
> After performing the configuration change, you should, run the
<A
HREF="deploymentconcerns.html#TESTSLONYSTATE"
>Section 5.1.1</A
> scripts in order to
validate that the cluster state remains in good order after this
change. </P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN1283"
>3.4.3. Failover</A
></H2
><A
NAME="AEN1285"
></A
><P
> If some more serious problem occurs on the
<SPAN
CLASS="QUOTE"
>"origin"</SPAN
> server, it may be necessary to <A
HREF="stmtfailover.html"
>SLONIK FAILOVER</A
> to a backup server. This is a highly
undesirable circumstance, as transactions <SPAN
CLASS="QUOTE"
>"committed"</SPAN
> on
the origin, but not applied to the subscribers, will be lost. You may
have reported these transactions as <SPAN
CLASS="QUOTE"
>"successful"</SPAN
> to
outside users. As a result, failover should be considered a
<SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>last resort</I
></SPAN
>. If the <SPAN
CLASS="QUOTE"
>"injured"</SPAN
>
origin server can be brought up to the point where it can limp along
long enough to do a controlled switchover, that is
<SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>greatly</I
></SPAN
> preferable.</P
><P
> <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> does not provide any automatic detection for failed
systems. Abandoning committed transactions is a business decision
that cannot be made by a database system. If someone wants to put the
commands below into a script executed automatically from the network
monitoring system, well ... it's <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>your</I
></SPAN
> data, and
it's <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>your</I
></SPAN
> failover policy. </P
><P
></P
><UL
><LI
><P
>The <A
HREF="slonik.html"
><SPAN
CLASS="APPLICATION"
>slonik</SPAN
></A
> command
</P><PRE
CLASS="PROGRAMLISTING"
>failover (id = 1, backup node = 2);</PRE
><P></P
><P
> causes node2 to assume the ownership (origin) of all sets that
have node1 as their current origin. If there should happen to be
additional nodes in the <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> cluster, all direct subscribers of
node1 are instructed that this is happening.
<SPAN
CLASS="APPLICATION"
>Slonik</SPAN
> will also query all direct
subscribers in order to determine out which node has the highest
replication status (<SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>e.g.</I
></SPAN
> - the latest committed
transaction) for each set, and the configuration will be changed in a
way that node2 first applies those final before actually allowing
write access to the tables.</P
><P
> In addition, all nodes that subscribed directly to node1 will
now use node2 as data provider for the set. This means that after the
failover command succeeded, no node in the entire replication setup
will receive anything from node1 any more.</P
><DIV
CLASS="NOTE"
><BLOCKQUOTE
CLASS="NOTE"
><P
><B
>Note: </B
> Note that in order for node 2 to be considered as a
candidate for failover, it must have been set up with the <A
HREF="stmtsubscribeset.html"
>SLONIK SUBSCRIBE SET</A
> option <TT
CLASS="COMMAND"
>forwarding =
yes</TT
>, which has the effect that replication log data is
collected in <A
HREF="table.sl-log-1.html"
>sl_log_1</A
>/<A
HREF="table.sl-log-2.html"
>sl_log_2</A
> on node 2. If replication log data is
<SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>not</I
></SPAN
> being collected, then failover to that node
is not possible. </P
><P
>Only nodes listed in the failover_candidates view
can be used as the failover target for an origin. This is discussed in further detail below.</P
></BLOCKQUOTE
></DIV
></LI
><LI
><P
> Reconfigure and restart the application (or
<SPAN
CLASS="APPLICATION"
>pgpool</SPAN
>) to cause it to reconnect to
node2.</P
></LI
><LI
><P
> Purge out the abandoned node </P
><P
> You will find, after the failover, that there are still a full
set of references to node 1 in <A
HREF="table.sl-node.html"
>sl_node</A
>, as well
as in referring tables such as <A
HREF="table.sl-confirm.html"
>sl_confirm</A
>;
since data in <A
HREF="table.sl-log-1.html"
>sl_log_1</A
>/<A
HREF="table.sl-log-2.html"
>sl_log_2</A
> is still present, <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> cannot
immediately purge out the node. </P
><P
> After the failover is complete and all nodes have been
reconfigured you can remove all remnants of node1's
configuration information with the <A
HREF="stmtdropnode.html"
>SLONIK DROP NODE</A
>
command:
</P><PRE
CLASS="PROGRAMLISTING"
>drop node (id = 1, event node = 2);</PRE
><P></P
><P
> Supposing the failure resulted from some catastrophic failure
of the hardware supporting node 1, there might be no
<SPAN
CLASS="QUOTE"
>"remains"</SPAN
> left to look at. If the failure was not
<SPAN
CLASS="QUOTE"
>"total"</SPAN
>, as might be the case if the node had to be
abandoned due to a network communications failure, you will find that
node 1 still <SPAN
CLASS="QUOTE"
>"imagines"</SPAN
> itself to be as it was before the
failure. See <A
HREF="failover.html#REBUILDNODE1"
>Section 3.4.7</A
> for more details on the
implications.</P
></LI
><LI
><P
> After performing the configuration change, you
should, as run the <A
HREF="deploymentconcerns.html#TESTSLONYSTATE"
>Section 5.1.1</A
>
scripts in order to validate that the cluster state remains in good
order after this change. </P
></LI
></UL
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="FAILOVERTARGETS"
>3.4.4. Failover Targets</A
></H2
><P
>An origin node can only be failed over to other nodes in the cluster that
are valid failover targets. The failover targets for a node must meet the
following conditions</P
><P
></P
><UL
><LI
><P
>The failover target must be a direct subscriber of all sets that
the failed node is an origin for</P
></LI
><LI
><P
>The failover target must have paths to all nodes that the
failed node has paths to</P
></LI
></UL
><P
>The view sl_failover_targets displays the valid failover targets for
each origin node. Clusters that have more than two nodes and would like
to have the option of using failover need to be setup in such a way that
valid failover targets exist for the various failure scenarios that
they wish to support.</P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="MULTIPLEFAILURES"
>3.4.5. Multiple Node Failures</A
></H2
><P
>If multiple nodes fail at the same time, maybe because an entire data-center has failed,
then all failed nodes should be passed to the failover command. If we consider a cluster where node 1 the origin of a set and provides a subscription to
node 2 and node 3 then node 2 provides a subscription to node 4, what should happen if both nodes 1 and 2 fail?
Slony can be told about the failed nodes with the following command
</P><PRE
CLASS="PROGRAMLISTING"
>FAILOVER (node=(id=1, backup node=3), node=(id=2, backup node=3));</PRE
><P>
This command requires that a paths exist between node 3 and 4. It will then redirect node 4 to receive the subscription from node 3.
</P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN1352"
>3.4.6. Automating <TT
CLASS="COMMAND"
> FAIL OVER </TT
></A
></H2
><A
NAME="AEN1355"
></A
><P
> If you do choose to automate <TT
CLASS="COMMAND"
>FAIL OVER </TT
>, it
is important to do so <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>carefully.</I
></SPAN
> You need to have
good assurance that the failed node is well and truly failed, and you
need to be able to assure that the failed node will not accidentally
return into service, thereby allowing there to be two nodes out there
able to respond in a <SPAN
CLASS="QUOTE"
>"master"</SPAN
> role. </P
><DIV
CLASS="NOTE"
><BLOCKQUOTE
CLASS="NOTE"
><P
><B
>Note: </B
> The problem here requiring that you <SPAN
CLASS="QUOTE"
>"shoot the
failed node in the head"</SPAN
> is not fundamentally about replication
or <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
>; <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> handles this all reasonably gracefully, as once
the node is marked as failed, the other nodes will <SPAN
CLASS="QUOTE"
>"shun"</SPAN
>
it, effectively ignoring it. The problem is instead with
<SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>your application.</I
></SPAN
> Supposing the failed node can
come back up sufficiently that it can respond to application requests,
<SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>that</I
></SPAN
> is likely to be a problem, and one that
hasn't anything to do with <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
>. The trouble is if there are two
databases that can respond as if they are <SPAN
CLASS="QUOTE"
>"master"</SPAN
>
systems. </P
></BLOCKQUOTE
></DIV
><P
> When failover occurs, there therefore needs to be a mechanism
to forcibly knock the failed node off the network in order to prevent
applications from getting confused. This could take place via having
an SNMP interface that does some combination of the following:</P
><P
></P
><UL
><LI
><P
> Turns off power on the failed server. </P
><P
> If care is not taken, the server may reappear when system
administrators power it up. </P
></LI
><LI
><P
> Modify firewall rules or other network configuration
to drop the failed server's IP address from the network. </P
><P
> If the server has multiple network interfaces, and therefore
multiple IP addresses, this approach allows the
<SPAN
CLASS="QUOTE"
>"application"</SPAN
> addresses to be dropped/deactivated, but
leave <SPAN
CLASS="QUOTE"
>"administrative"</SPAN
> addresses open so that the server
would remain accessible to system administrators. </P
></LI
></UL
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="REBUILDNODE1"
>3.4.7. After Failover, Reconfiguring
Former Origin</A
></H2
><A
NAME="AEN1383"
></A
><P
> What happens to the failed node will depend somewhat on the
nature of the catastrophe that lead to needing to fail over to another
node. If the node had to be abandoned because of physical destruction
of its disk storage, there will likely not be anything of interest
left. On the other hand, a node might be abandoned due to the failure
of a network connection, in which case the former
<SPAN
CLASS="QUOTE"
>"provider"</SPAN
> can appear be functioning perfectly well.
Nonetheless, once communications are restored, the fact of the
<TT
CLASS="COMMAND"
>FAIL OVER</TT
> makes it mandatory that the failed node
be abandoned. </P
><P
> After the above failover, the data stored on node 1 will
rapidly become increasingly out of sync with the rest of the nodes,
and must be treated as corrupt. Therefore, the only way to get node 1
back and transfer the origin role back to it is to rebuild it from
scratch as a subscriber, let it catch up, and then follow the
switchover procedure.</P
><P
> A good reason <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>not</I
></SPAN
> to do this automatically
is the fact that important updates (from a
<SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>business</I
></SPAN
> perspective) may have been
<TT
CLASS="COMMAND"
>commit</TT
>ted on the failing system. You probably want
to analyze the last few transactions that made it into the failed node
to see if some of them need to be reapplied on the <SPAN
CLASS="QUOTE"
>"live"</SPAN
>
cluster. For instance, if someone was entering bank deposits
affecting customer accounts at the time of failure, you wouldn't want
to lose that information.</P
><DIV
CLASS="WARNING"
><P
></P
><TABLE
CLASS="WARNING"
BORDER="1"
WIDTH="100%"
><TR
><TD
ALIGN="CENTER"
><B
>Warning</B
></TD
></TR
><TR
><TD
ALIGN="LEFT"
><P
> It has been observed that there can be some very
confusing results if a node is <SPAN
CLASS="QUOTE"
>"failed"</SPAN
> due to a
persistent network outage as opposed to failure of data storage. In
such a scenario, the <SPAN
CLASS="QUOTE"
>"failed"</SPAN
> node has a database in
perfectly fine form; it is just that since it was cut off, it
<SPAN
CLASS="QUOTE"
>"screams in silence."</SPAN
> </P
><P
> If the network connection is repaired, that node could
reappear, and as far as <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>its</I
></SPAN
> configuration is
concerned, all is well, and it should communicate with the rest of its
<SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> cluster. </P
><P
> In <SPAN
CLASS="emphasis"
><I
CLASS="EMPHASIS"
>fact</I
></SPAN
>, the only confusion taking place
is on that node. The other nodes in the cluster are not confused at
all; they know that this node is <SPAN
CLASS="QUOTE"
>"dead,"</SPAN
> and that they
should ignore it. But there is not a way to know this by looking at
the <SPAN
CLASS="QUOTE"
>"failed"</SPAN
> node.</P
><P
> This points back to the design point that <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> is not a
network monitoring tool. You need to have clear methods of
communicating to applications and users what database hosts are to be
used. If those methods are lacking, adding replication to the mix
will worsen the potential for confusion, and failover will be a point
at which there is enormous potential for confusion. </P
></TD
></TR
></TABLE
></DIV
><P
> If the database is very large, it may take many hours to
recover node1 as a functioning <SPAN
CLASS="PRODUCTNAME"
>Slony-I</SPAN
> node; that is another reason
to consider failover as an undesirable <SPAN
CLASS="QUOTE"
>"final
resort."</SPAN
></P
></DIV
><DIV
CLASS="SECT2"
><H2
CLASS="SECT2"
><A
NAME="AEN1411"
>3.4.8. Planning for Failover</A
></H2
><P
> <A
HREF="failover.html"
> Failover </A
> policies
should be planned for ahead of time. </P
><P
> Most pointedly, any node that is expected to be a failover
target must have its subscription(s) set up with the option
<TT
CLASS="COMMAND"
>FORWARD = YES</TT
>. Otherwise, that node is not a
candidate for being promoted to origin node. </P
><P
> This may simply involve thinking about what the priority lists
should be of what should fail to what, as opposed to trying to
automate it. But knowing what to do ahead of time cuts down on the
number of mistakes made.</P
><P
> At Afilias, a variety of internal [<SPAN
CLASS="CITATION"
>The 3AM Unhappy
DBA's Guide to...</SPAN
>] guides have been created to provide
checklists of what to do when certain <SPAN
CLASS="QUOTE"
>"unhappy"</SPAN
> events
take place. This sort of material is highly specific to the
environment and the set of applications running there, so you would
need to generate your own such documents. This is one of the vital
components of any disaster recovery preparations.</P
></DIV
></DIV
><DIV
CLASS="NAVFOOTER"
><HR
ALIGN="LEFT"
WIDTH="100%"><TABLE
SUMMARY="Footer navigation table"
WIDTH="100%"
BORDER="0"
CELLPADDING="0"
CELLSPACING="0"
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
><A
HREF="ddlchanges.html"
ACCESSKEY="P"
>Prev</A
></TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="index.html"
ACCESSKEY="H"
>Home</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
><A
HREF="advanced.html"
ACCESSKEY="N"
>Next</A
></TD
></TR
><TR
><TD
WIDTH="33%"
ALIGN="left"
VALIGN="top"
>Database Schema Changes (DDL)</TD
><TD
WIDTH="34%"
ALIGN="center"
VALIGN="top"
><A
HREF="administration.html"
ACCESSKEY="U"
>Up</A
></TD
><TD
WIDTH="33%"
ALIGN="right"
VALIGN="top"
>Advanced Concepts</TD
></TR
></TABLE
></DIV
></BODY
></HTML
>
|