docs/programmer_reference/transapp_hotfail.html

   1 <?xml version="1.0" encoding="UTF-8" standalone="no"?>
   2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
   3 <html xmlns="http://www.w3.org/1999/xhtml">
   4   <head>
   5     <meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
   6     <title>Hot failover</title>
   7     <link rel="stylesheet" href="gettingStarted.css" type="text/css" />
   8     <meta name="generator" content="DocBook XSL Stylesheets V1.73.2" />
   9     <link rel="start" href="index.html" title="Berkeley DB Programmer's Reference Guide" />
  10     <link rel="up" href="transapp.html" title="Chapter 11.  Berkeley DB Transactional Data Store Applications" />
  11     <link rel="prev" href="transapp_recovery.html" title="Recovery procedures" />
  12     <link rel="next" href="transapp_journal.html" title="Using Recovery on Journaling Filesystems" />
  13   </head>
  14   <body>
  15     <div xmlns="" class="navheader">
  16       <div class="libver">
  17         <p>Library Version 11.2.5.3</p>
  18       </div>
  19       <table width="100%" summary="Navigation header">
  20         <tr>
  21           <th colspan="3" align="center">Hot failover</th>
  22         </tr>
  23         <tr>
  24           <td width="20%" align="left"><a accesskey="p" href="transapp_recovery.html">Prev</a> </td>
  25           <th width="60%" align="center">Chapter 11.
  26                 Berkeley DB Transactional Data Store Applications
  27         </th>
  28           <td width="20%" align="right"> <a accesskey="n" href="transapp_journal.html">Next</a></td>
  29         </tr>
  30       </table>
  31       <hr />
  32     </div>
  33     <div class="sect1" lang="en" xml:lang="en">
  34       <div class="titlepage">
  35         <div>
  36           <div>
  37             <h2 class="title" style="clear: both"><a id="transapp_hotfail"></a>Hot failover</h2>
  38           </div>
  39         </div>
  40       </div>
  41       <p>
  42         For some applications, it may be useful to periodically snapshot
  43         the database environment for use as a hot failover should the
  44         primary system fail.  The following steps can be taken to keep a
  45         backup environment in close synchrony with an active environment.
  46         The active environment is entirely unaffected by these procedures,
  47         and both read and write operations are allowed during all steps
  48         described here.
  49     </p>
  50       <p>
  51         The procedure described here is not compatible with the concurrent
  52         use of the transactional bulk insert optimization (transactions
  53         started with the <a href="../api_reference/C/txnbegin.html#txnbegin_DB_TXN_BULK" class="olink">DB_TXN_BULK</a> flag). After the bulk optimization
  54         is used, the archive must be created again from scratch starting
  55         with step 1.
  56     </p>
  57       <p>
  58         The <a href="../api_reference/C/db_hotbackup.html" class="olink">db_hotbackup</a> utility is the preferred way to automate generating a hot
  59         failover system.  The first step is to run <a href="../api_reference/C/db_hotbackup.html" class="olink">db_hotbackup</a> utility without the
  60         <span class="bold"><strong>-u</strong></span> flag.  This will create hot
  61         backup copy of the databases in your environment.  After that point
  62         periodically running the <a href="../api_reference/C/db_hotbackup.html" class="olink">db_hotbackup</a> utility with the
  63         <span class="bold"><strong>-u</strong></span> flag will copy the new
  64         log files and run recovery on the backup copy to bring it current
  65         with the primary environment.
  66     </p>
  67       <p>
  68         Note that you can also create your own hot backup solution using
  69         the <a href="../api_reference/C/envbackup.html" class="olink">DB_ENV-&gt;backup()</a> or <a href="../api_reference/C/envdbbackup.html" class="olink">DB_ENV-&gt;dbbackup()</a> methods.
  70     </p>
  71       <p>
  72         To implement your own hot fail over system, the steps below can be
  73         followed. However, care should be taken on non-UNIX based systems
  74         when copying the database files to be sure that they are either
  75         quiescent, or that either the <a href="../api_reference/C/envbackup.html" class="olink">DB_ENV-&gt;backup()</a> or <a href="../api_reference/C/db_copy.html" class="olink">db_copy()</a> routine is
  76         used to ensure atomic reads of the database pages.
  77     </p>
  78       <div class="orderedlist">
  79         <ol type="1">
  80           <li>
  81             <p>
  82                 Run the <a href="../api_reference/C/db_archive.html" class="olink">db_archive</a> utility with the <span class="bold"><strong>-s</strong></span> option in the active environment
  83                 to identify all of the active environment's database files, and
  84                 copy them to the backup directory.
  85             </p>
  86             <p>
  87                 If the database files are stored in a separate directory from
  88                 the other Berkeley DB files, it will be simpler (and much
  89                 faster!) to copy the directory itself instead of the individual
  90                 files (see <a href="../api_reference/C/envadd_data_dir.html" class="olink">DB_ENV-&gt;add_data_dir()</a> for additional information).
  91             </p>
  92             <div class="note" style="margin-left: 0.5in; margin-right: 0.5in;">
  93               <h3 class="title">Note</h3>
  94               <p>
  95                     If any of the database files did not have an open <a href="../api_reference/C/db.html" class="olink">DB</a>
  96                     handle during the lifetime of the current log files,
  97                     the <a href="../api_reference/C/db_archive.html" class="olink">db_archive</a> utility will not list them in its output. This
  98                     is another reason it may be simpler to use a separate
  99                     database file directory and copy the entire directory
 100                     instead of archiving only the files listed by the
 101                     <a href="../api_reference/C/db_archive.html" class="olink">db_archive</a> utility.
 102                 </p>
 103             </div>
 104           </li>
 105           <li>
 106             Remove all existing log files from the backup directory.
 107         </li>
 108           <li>
 109             Run the <a href="../api_reference/C/db_archive.html" class="olink">db_archive</a> utility with the <span class="bold"><strong>-l</strong></span>
 110             option in the active environment to identify all of the active
 111             environment's log files, and copy them to the backup directory.
 112         </li>
 113           <li>
 114             Run the <a href="../api_reference/C/db_recover.html" class="olink">db_recover</a> utility with the <span class="bold"><strong>-c</strong></span>
 115             option in the backup directory to catastrophically recover the
 116             copied environment.
 117         </li>
 118         </ol>
 119       </div>
 120       <p>
 121         Steps 2, 3 and 4 may be repeated as often as you like.  If Step 1
 122         (the initial copy of the database files) is repeated, then Steps 2,
 123         3 and 4 <span class="bold"><strong>must</strong></span> be performed at least
 124         once in order to ensure a consistent database environment
 125         snapshot.
 126     </p>
 127       <p>
 128         These procedures must be integrated with your other archival
 129         procedures, of course.  If you are periodically removing log files
 130         from your active environment, you must be sure to copy them to the
 131         backup directory before removing them from the active directory.
 132         Not copying a log file to the backup directory and subsequently
 133         running recovery with it present may leave the backup snapshot of
 134         the environment corrupted.  A simple way to ensure this never
 135         happens is to archive the log files in Step 2 as you remove them
 136         from the backup directory, and move inactive log files from your
 137         active environment into your backup directory (rather than copying
 138         them), in Step 3.  The following steps describe this procedure in
 139         more detail:
 140     </p>
 141       <div class="orderedlist">
 142         <ol type="1">
 143           <li>
 144             Run the <a href="../api_reference/C/db_archive.html" class="olink">db_archive</a> utility with the <span class="bold"><strong>-s</strong></span>
 145             option in the active environment to identify all of the active
 146             environment's database files, and copy them to the backup
 147             directory.
 148         </li>
 149           <li>
 150             Archive all existing log files from the backup directory, moving them
 151             to a backup device such as CD-ROM, alternate disk, or tape.
 152         </li>
 153           <li>
 154             Run the <a href="../api_reference/C/db_archive.html" class="olink">db_archive</a> utility (without any option) in the active environment
 155             to identify all of the log files in the active environment that are
 156             no longer in use, and <span class="bold"><strong>move</strong></span> them to
 157             the backup directory.
 158         </li>
 159           <li>
 160             Run the <a href="../api_reference/C/db_archive.html" class="olink">db_archive</a> utility with the <span class="bold"><strong>-l</strong></span>
 161             option in the active environment to identify all of the remaining
 162             log files in the active environment, and <span class="bold"><strong>copy</strong></span> the log files to the backup
 163             directory.
 164         </li>
 165           <li>
 166             Run the <a href="../api_reference/C/db_recover.html" class="olink">db_recover</a> utility with the <span class="bold"><strong>-c</strong></span>
 167             option in the backup directory to catastrophically recover the
 168             copied environment.
 169         </li>
 170         </ol>
 171       </div>
 172       <p>
 173         As before, steps 2, 3, 4 and 5 may be repeated as often as you
 174         like.  If Step 1 (the initial copy of the database files) is
 175         repeated, then Steps 2 through 5
 176         <span class="bold"><strong>must</strong></span> be performed at least once in
 177         order to ensure a consistent database environment snapshot.
 178     </p>
 179     </div>
 180     <div class="navfooter">
 181       <hr />
 182       <table width="100%" summary="Navigation footer">
 183         <tr>
 184           <td width="40%" align="left"><a accesskey="p" href="transapp_recovery.html">Prev</a> </td>
 185           <td width="20%" align="center">
 186             <a accesskey="u" href="transapp.html">Up</a>
 187           </td>
 188           <td width="40%" align="right"> <a accesskey="n" href="transapp_journal.html">Next</a></td>
 189         </tr>
 190         <tr>
 191           <td width="40%" align="left" valign="top">Recovery procedures </td>
 192           <td width="20%" align="center">
 193             <a accesskey="h" href="index.html">Home</a>
 194           </td>
 195           <td width="40%" align="right" valign="top"> Using Recovery on Journaling Filesystems</td>
 196         </tr>
 197       </table>
 198     </div>
 199   </body>
 200 </html>