<?xml version="1.0" encoding="utf-8" ?> <!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" xml:lang="en" lang="en"> <head> <meta http-equiv="Content-Type" content="text/html; charset=utf-8" /> <meta name="generator" content="Docutils 0.5: http://docutils.sourceforge.net/" /> <title>Xapian Administrator's Guide</title> <style type="text/css"> /* :Author: David Goodger (goodger@python.org) :Id: $Id: html4css1.css 5196 2007-06-03 20:25:28Z wiemann $ :Copyright: This stylesheet has been placed in the public domain. Default cascading style sheet for the HTML output of Docutils. See http://docutils.sf.net/docs/howto/html-stylesheets.html for how to customize this style sheet. */ /* used to remove borders from tables and images */ .borderless, table.borderless td, table.borderless th { border: 0 } table.borderless td, table.borderless th { /* Override padding for "table.docutils td" with "! important". The right padding separates the table cells. */ padding: 0 0.5em 0 0 ! important } .first { /* Override more specific margin styles with "! important". */ margin-top: 0 ! important } .last, .with-subtitle { margin-bottom: 0 ! important } .hidden { display: none } a.toc-backref { text-decoration: none ; color: black } blockquote.epigraph { margin: 2em 5em ; } dl.docutils dd { margin-bottom: 0.5em } /* Uncomment (and remove this text!) to get bold-faced definition list terms dl.docutils dt { font-weight: bold } */ div.abstract { margin: 2em 5em } div.abstract p.topic-title { font-weight: bold ; text-align: center } div.admonition, div.attention, div.caution, div.danger, div.error, div.hint, div.important, div.note, div.tip, div.warning { margin: 2em ; border: medium outset ; padding: 1em } div.admonition p.admonition-title, div.hint p.admonition-title, div.important p.admonition-title, div.note p.admonition-title, div.tip p.admonition-title { font-weight: bold ; font-family: sans-serif } div.attention p.admonition-title, div.caution p.admonition-title, div.danger p.admonition-title, div.error p.admonition-title, div.warning p.admonition-title { color: red ; font-weight: bold ; font-family: sans-serif } /* Uncomment (and remove this text!) to get reduced vertical space in compound paragraphs. div.compound .compound-first, div.compound .compound-middle { margin-bottom: 0.5em } div.compound .compound-last, div.compound .compound-middle { margin-top: 0.5em } */ div.dedication { margin: 2em 5em ; text-align: center ; font-style: italic } div.dedication p.topic-title { font-weight: bold ; font-style: normal } div.figure { margin-left: 2em ; margin-right: 2em } div.footer, div.header { clear: both; font-size: smaller } div.line-block { display: block ; margin-top: 1em ; margin-bottom: 1em } div.line-block div.line-block { margin-top: 0 ; margin-bottom: 0 ; margin-left: 1.5em } div.sidebar { margin: 0 0 0.5em 1em ; border: medium outset ; padding: 1em ; background-color: #ffffee ; width: 40% ; float: right ; clear: right } div.sidebar p.rubric { font-family: sans-serif ; font-size: medium } div.system-messages { margin: 5em } div.system-messages h1 { color: red } div.system-message { border: medium outset ; padding: 1em } div.system-message p.system-message-title { color: red ; font-weight: bold } div.topic { margin: 2em } h1.section-subtitle, h2.section-subtitle, h3.section-subtitle, h4.section-subtitle, h5.section-subtitle, h6.section-subtitle { margin-top: 0.4em } h1.title { text-align: center } h2.subtitle { text-align: center } hr.docutils { width: 75% } img.align-left { clear: left } img.align-right { clear: right } ol.simple, ul.simple { margin-bottom: 1em } ol.arabic { list-style: decimal } ol.loweralpha { list-style: lower-alpha } ol.upperalpha { list-style: upper-alpha } ol.lowerroman { list-style: lower-roman } ol.upperroman { list-style: upper-roman } p.attribution { text-align: right ; margin-left: 50% } p.caption { font-style: italic } p.credits { font-style: italic ; font-size: smaller } p.label { white-space: nowrap } p.rubric { font-weight: bold ; font-size: larger ; color: maroon ; text-align: center } p.sidebar-title { font-family: sans-serif ; font-weight: bold ; font-size: larger } p.sidebar-subtitle { font-family: sans-serif ; font-weight: bold } p.topic-title { font-weight: bold } pre.address { margin-bottom: 0 ; margin-top: 0 ; font-family: serif ; font-size: 100% } pre.literal-block, pre.doctest-block { margin-left: 2em ; margin-right: 2em } span.classifier { font-family: sans-serif ; font-style: oblique } span.classifier-delimiter { font-family: sans-serif ; font-weight: bold } span.interpreted { font-family: sans-serif } span.option { white-space: nowrap } span.pre { white-space: pre } span.problematic { color: red } span.section-subtitle { /* font-size relative to parent (h1..h6 element) */ font-size: 80% } table.citation { border-left: solid 1px gray; margin-left: 1px } table.docinfo { margin: 2em 4em } table.docutils { margin-top: 0.5em ; margin-bottom: 0.5em } table.footnote { border-left: solid 1px black; margin-left: 1px } table.docutils td, table.docutils th, table.docinfo td, table.docinfo th { padding-left: 0.5em ; padding-right: 0.5em ; vertical-align: top } table.docutils th.field-name, table.docinfo th.docinfo-name { font-weight: bold ; text-align: left ; white-space: nowrap ; padding-left: 0 } h1 tt.docutils, h2 tt.docutils, h3 tt.docutils, h4 tt.docutils, h5 tt.docutils, h6 tt.docutils { font-size: 100% } ul.auto-toc { list-style-type: none } </style> </head> <body> <div class="document" id="xapian-administrator-s-guide"> <h1 class="title">Xapian Administrator's Guide</h1> <!-- Copyright (C) 2006 Lemur Consulting Ltd --> <!-- Copyright (C) 2007,2008,2009,2010,2011 Olly Betts --> <!-- FIXME: Once brass settles down, update this for brass --> <div class="contents topic" id="table-of-contents"> <p class="topic-title first">Table of contents</p> <ul class="simple"> <li><a class="reference internal" href="#introduction" id="id1">Introduction</a></li> <li><a class="reference internal" href="#databases" id="id2">Databases</a><ul> <li><a class="reference internal" href="#atomic-modifications" id="id3">Atomic modifications</a></li> <li><a class="reference internal" href="#single-writer-multiple-reader" id="id4">Single writer, multiple reader</a></li> <li><a class="reference internal" href="#revision-numbers" id="id5">Revision numbers</a></li> <li><a class="reference internal" href="#network-file-systems" id="id6">Network file systems</a></li> <li><a class="reference internal" href="#which-database-format-to-use" id="id7">Which database format to use?</a></li> <li><a class="reference internal" href="#can-i-put-other-files-in-the-database-directory" id="id8">Can I put other files in the database directory?</a></li> </ul> </li> <li><a class="reference internal" href="#backup-strategies" id="id9">Backup Strategies</a><ul> <li><a class="reference internal" href="#summary" id="id10">Summary</a></li> <li><a class="reference internal" href="#detail" id="id11">Detail</a></li> </ul> </li> <li><a class="reference internal" href="#inspecting-a-database" id="id12">Inspecting a database</a></li> <li><a class="reference internal" href="#database-maintenance" id="id13">Database maintenance</a><ul> <li><a class="reference internal" href="#compacting-a-database" id="id14">Compacting a database</a></li> <li><a class="reference internal" href="#merging-databases" id="id15">Merging databases</a></li> <li><a class="reference internal" href="#checking-database-integrity" id="id16">Checking database integrity</a></li> <li><a class="reference internal" href="#converting-a-pre-1-1-4-chert-database-to-a-chert-database" id="id17">Converting a pre-1.1.4 chert database to a chert database</a></li> <li><a class="reference internal" href="#converting-a-flint-database-to-a-chert-database" id="id18">Converting a flint database to a chert database</a></li> <li><a class="reference internal" href="#converting-a-quartz-database-to-a-flint-database" id="id19">Converting a quartz database to a flint database</a></li> <li><a class="reference internal" href="#converting-a-0-9-x-flint-database-to-work-with-1-0-y" id="id20">Converting a 0.9.x flint database to work with 1.0.y</a></li> </ul> </li> </ul> </div> <div class="section" id="introduction"> <h1><a class="toc-backref" href="#id1">Introduction</a></h1> <p>This document is intended to provide general hints, tips and advice to administrators of Xapian systems. It assumes that you have installed Xapian on your system, and are familiar with the basics of creating and searching Xapian databases.</p> <p>The intended audience is system administrators who need to be able to perform general management of a Xapian database, including tasks such as taking backups and optimising performance. It may also be useful introductory reading for Xapian application developers.</p> <p>The document is up-to-date for Xapian version 1.2.10.</p> </div> <div class="section" id="databases"> <h1><a class="toc-backref" href="#id2">Databases</a></h1> <p>Xapian databases hold all the information needed to perform searches in a set of tables. The following tables always exist:</p> <blockquote> <ul class="simple"> <li>A posting list table, which holds a list of all the documents indexed by each term in the database, and (for chert) also chunked streams of the values in each value slot.</li> <li>A record table, which holds the document data associated with each document in the database.</li> <li>A termlist table, which holds a list of all the terms which index each document, and (for chert) also the value slots used in each document.</li> </ul> </blockquote> <p>And the following optional tables exist only when there is data to store in them (in 1.0.1 and earlier, the position and value tables were always created even if empty; spelling and synonym tables were new in 1.0.2):</p> <blockquote> <ul class="simple"> <li>A position list table, which holds a list of all the word positions in each document which each term occurs at.</li> <li>A value table, which holds the "values" (used for sorting, collapsing, and other match-time calculations) associated with each document in the database (only for flint - newer backends store values in the postlist and termlist tables).</li> <li>A spelling table, which holds data for suggesting spelling corrections.</li> <li>A synonym table, which holds a synonym dictionary.</li> </ul> </blockquote> <p>Each of the tables is held in a separate file, allowing an administrator to see how much data is being used for each of the above purposes. It is not always necessary to fully populate these tables: for example, if phrase searches are never going to be performed on the database, it is not necessary to store any positionlist information.</p> <p>If you look at a Xapian database, you will see that each of these tables actually uses 2 or 3 files. For example, for a "chert" format database the termlist table is stored in the files "termlist.baseA", "termlist.baseB" and "termlist.DB".</p> <p>The ".DB" file actually stores the data, and is structured as a tree of blocks, which have a default size of 8KB (though this can be set, either through the Xapian API, or with some of the tools detailed later in this document).</p> <p>The ".baseA" and ".baseB" files are used to keep track of where to start looking for data in the ".DB" file (the root of the tree), and which blocks are in use. Often only one of the ".baseA" and ".baseB" files will be present; each of these files refers to a revision of the database, and there may be more than one valid revision of the database stored in the ".DB" file at once.</p> <p>Changing the blocksize may have performance implications, but it is hard to tell whether these will be positive or negative for a particular combination of hardware and software without doing some profiling.</p> <div class="section" id="atomic-modifications"> <h2><a class="toc-backref" href="#id3">Atomic modifications</a></h2> <p>Xapian ensures that all modifications to its database are performed atomically. This means that:</p> <blockquote> <ul class="simple"> <li>From the point of view of a separate process (or a separate database object in the same process) reading the database, all modifications made to a database are invisible until the modifications is committed.</li> <li>The database on disk is always in a consistent state.</li> <li>If the system is interrupted during a modification, the database should always be left in a valid state. This applies even if the power is cut unexpectedly, as long as the disk does not become corrupted due to hardware failure.</li> </ul> </blockquote> <p>Committing a modification requires several calls to the operating system to make it flush any cached modifications to the database to disk. This is to ensure that if the system fails at any point, the database is left in a consistent state. Of course, this is a fairly slow process (since the system has to wait for the disk to physically write the data), so grouping many changes together will speed up the throughput considerably.</p> <p>Many modifications can be explicitly grouped into a single transaction, so that lots of changes are applied at once. Even if an application doesn't explicitly protect modifications to the database using transactions, Xapian will group modifications into transactions, applying the modifications in batches.</p> <p>Note that it is not currently possible to extend Xapian's transactions to cover multiple databases, or to link them with transactions in external systems, such as an RDBMS.</p> <p>Finally, note that it is possible to compile Xapian such that it doesn't make modifications in an atomic manner, in order to build very large databases more quickly (search the Xapian mailing list archives for "DANGEROUS" mode for more details). This isn't yet integrated into standard builds of Xapian, but may be in future, if appropriate protections can be incorporated.</p> </div> <div class="section" id="single-writer-multiple-reader"> <h2><a class="toc-backref" href="#id4">Single writer, multiple reader</a></h2> <p>Xapian implements a "single writer, multiple reader" model. This means that, at any given instant, there is only permitted to be a single object modifying a database, but there may (simultaneously) be many objects reading the database at once.</p> <p>Xapian enforces this restriction using by having a writer lock the database. Each Xapian database directory contains a lock file named <tt class="docutils literal"><span class="pre">flintlock</span></tt> (we've kept the same name as flint used, since the locking technique is the same).</p> <p>This lock-file will always exist, but will be locked using <tt class="docutils literal"><span class="pre">fcntl()</span></tt> when the database is open for writing. Because of the semantics of <tt class="docutils literal"><span class="pre">fcntl()</span></tt> locking, for each WritableDatabase opened we spawn a child process to hold the lock, which then exec-s <tt class="docutils literal"><span class="pre">cat</span></tt>, so you will see a <tt class="docutils literal"><span class="pre">cat</span></tt> subprocess of any writer process in the output of <tt class="docutils literal"><span class="pre">ps</span></tt>, <tt class="docutils literal"><span class="pre">top</span></tt>, etc.</p> <p>If a writer exits without being given a chance to clean up (for example, if the application holding the writer is killed), the <tt class="docutils literal"><span class="pre">fcntl()</span></tt> lock will be automatically released by the operating system. Under Microsoft Windows, we use a different locking technique which doesn't require a child process, but also means the lock is released automatically when the writing process exits.</p> </div> <div class="section" id="revision-numbers"> <h2><a class="toc-backref" href="#id5">Revision numbers</a></h2> <p>Xapian databases contain a revision number. This is essentially a count of the number of modifications since the database was created, and is needed to implement the atomic modification functionality. It is stored as a 32 bit integer, so there is a chance that a very frequently updated database could cause this to overflow. The consequence of such an overflow would be to throw an exception reporting that the database has run out of revision numbers.</p> <p>This isn't likely to be a practical problem, since it would take nearly a year for a database to reach this limit if 100 modifications were committed every second, and no normal Xapian system will commit more than once every few seconds. However, if you are concerned, you can use the <tt class="docutils literal"><span class="pre">xapian-compact</span></tt> tool to make a fresh copy of the database with the revision number set to 1.</p> <p>The revision number of each table can be displayed by the <tt class="docutils literal"><span class="pre">xapian-check</span></tt> tool.</p> </div> <div class="section" id="network-file-systems"> <h2><a class="toc-backref" href="#id6">Network file systems</a></h2> <p>Xapian should work correctly over a network file system. However, there are various potential issues with such file systems, so we recommend extensive testing of your particular network file system before deployment.</p> <p>Be warned that Xapian is heavily I/O dependent, and therefore performance over a network file system is likely to be slow unless you've got a very well tuned setup.</p> <p>Xapian needs to be able to lock a file in a database directory when modifications are being performed. On some network files systems (e.g., NFS) this requires a lock daemon to be running.</p> </div> <div class="section" id="which-database-format-to-use"> <h2><a class="toc-backref" href="#id7">Which database format to use?</a></h2> <p>As of release 1.2.0, you should generally use the chert format (which is now the default).</p> <p>Support for the pre-1.0 quartz format (deprecated in 1.0) was removed in 1.1.0. See below for how to convert a quartz database to a flint one.</p> <p>The flint backend (the default for 1.0) is still supported by 1.2.x, but deprecated - only use it if you already have flint databases; and plan to migrate away.</p> <p>There's also a development backend called brass. The main distinguishing feature of this is that the format may change incompatibly from time to time. It passes Xapian's extensive testsuite, but has seen less real world use than chert.</p> </div> <div class="section" id="can-i-put-other-files-in-the-database-directory"> <h2><a class="toc-backref" href="#id8">Can I put other files in the database directory?</a></h2> <p>If you wish to store meta-data or other information relating to the Xapian database, it is reasonable to wish to put this in files inside the Xapian database directory, for neatness. For example, you might wish to store a list of the prefixes you've applied to terms for specific fields in the database.</p> <p>Current Xapian backends don't do anything which will break this technique, so as long as you don't choose a filename that Xapian uses itself, there should be no problems. However, be aware that new versions of Xapian may use new files in the database directory, and it is also possible that new backend formats may not be compatible with the technique (e.g., it is possible that a future backend could store its entire database in a single file, not in a directory).</p> </div> </div> <div class="section" id="backup-strategies"> <h1><a class="toc-backref" href="#id9">Backup Strategies</a></h1> <div class="section" id="summary"> <h2><a class="toc-backref" href="#id10">Summary</a></h2> <blockquote> <ul class="simple"> <li>The simplest way to perform a backup is to temporarily halt modifications, take a copy of all files in the database directory, and then allow modifications to resume. Read access can continue while a backup is being taken.</li> <li>If you have a filesystem which allows atomic snapshots to be taken of directories (such as an LVM filesystem), an alternative strategy is to take a snapshot and simply copy all the files in the database directory to the backup medium. Such a copy will always be a valid database.</li> <li>Progressive backups are not easily possible; modifications are typically spread throughout the database files.</li> </ul> </blockquote> </div> <div class="section" id="detail"> <h2><a class="toc-backref" href="#id11">Detail</a></h2> <p>Even though Xapian databases are often automatically generated from source data which is stored in a reliable manner, it is usually desirable to keep backups of Xapian databases being run in production environments. This is particularly important in systems with high-availability requirements, since re-building a Xapian database from scratch can take many hours. It is also important in the case where the data stored in the database cannot easily be recovered from external sources.</p> <p>Xapian databases are managed such that at any instant in time, there is at least one valid revision of the database written to disk (and if there are multiple valid revisions, Xapian will always open the most recent). Therefore, if it is possible to take an instantaneous snapshot of all the database files (for example, on an LVM filesystem), this snapshot is suitable for copying to a backup medium. Note that it is not sufficient to take a snapshot of each database file in turn - the snapshot must be across all database files simultaneously. Otherwise, there is a risk that the snapshot could contain database files from different revisions.</p> <p>If it is not possible to take an instantaneous snapshot, the best backup strategy is simply to ensure that no modifications are committed during the backup procedure. While the simplest way to implement this may be to stop whatever processes are used to modify the database, and ensure that they close the database, it is not actually necessary to ensure that no writers are open on the database; it is enough to ensure that no writer makes any modification to the database.</p> <p>Because a Xapian database can contain more than one valid revision of the database, it is actually possible to allow a limited number of modifications to be performed while a backup copy is being made, but this is tricky and we do not recommend relying on it. Future versions of Xapian are likely to support this better, by allowing the current revision of a database to be preserved while modifications continue.</p> <p>Progressive backups are not recommended for Xapian databases: Xapian database files are block-structured, and modifications are spread throughout the /database file. Therefore, a progressive backup tool will not be able to take a backup by storing only the new parts of the database. Modifications will normally be so extensive that most parts of the database have been modified, however, if only a small number of modifications have been made, a binary diff algorithm might make a usable progressive backup tool.</p> </div> </div> <div class="section" id="inspecting-a-database"> <h1><a class="toc-backref" href="#id12">Inspecting a database</a></h1> <p>When designing an indexing strategy, it is often useful to be able to check the contents of the database. Xapian includes a simple command-line program, "delve", to allow this.</p> <p>For example, to display the list of terms in document "1" of the database "foo", use:</p> <pre class="literal-block"> delve foo -r 1 </pre> <p>It is also possible to perform simple searches of a database. Xapian includes another simple command-line program, "quest", to support this. "quest" is only able to search for un-prefixed terms, the query string must be quoted to protect it from the shell. To search the database "foo" for the phrase "hello world", use:</p> <pre class="literal-block"> quest -d foo '"hello world"' </pre> <p>If you have installed the "Omega" CGI application built on Xapian, this can also be used with the built-in "godmode" template to provide a web-based interface for browsing a database. See Omega's documentation for more details on this.</p> </div> <div class="section" id="database-maintenance"> <h1><a class="toc-backref" href="#id13">Database maintenance</a></h1> <div class="section" id="compacting-a-database"> <h2><a class="toc-backref" href="#id14">Compacting a database</a></h2> <p>Xapian databases normally have some spare space in each block to allow new information to be efficiently slotted into the database. However, the smaller a database is, the faster it can be searched, so if there aren't expected to be many further modifications, it can be desirable to compact the database.</p> <p>Xapian includes a tool called "xapian-compact" for compacting databases. This tool makes a copy of a database, and takes advantage of the sorted nature of the source Xapian database to write the database out without leaving spare space for future modifications. This can result in a large space saving.</p> <p>The downside of compaction is that future modifications may take a little longer, due to needing to reorganise the database to make space for them. However, modifications are still possible, and if many modifications are made, the database will gradually develop spare space.</p> <p>There's an option ("-F") to perform a "fuller" compaction. This option compacts the database as much as possible, but it violates the design of the Btree format slightly to achieve this, so it is not recommended if further modifications are at all likely in future. If you do need to modify a "fuller" compacted database, we recommend you run xapian-compact on it without "-F" first.</p> <p>While taking a copy of the database, it is also possible to change the blocksize. If you wish to profile search speed with different blocksizes, this is the recommended way to generate the different databases (but remember to compact the original database as well, for a fair comparison).</p> </div> <div class="section" id="merging-databases"> <h2><a class="toc-backref" href="#id15">Merging databases</a></h2> <p>When building an index for a very large amount of data, it can be desirable to index the data in smaller chunks (perhaps on separate machines), and then merge the chunks together into a single database. This can be performed using the "xapian-compact" tool, simply by supplying several source database paths.</p> <p>Normally, merging works by reading the source databases in parallel, and writing the contents in sorted order to the destination database. This will work most efficiently if excessive disk seeking can be avoided; if you have several disks, it may be worth placing the source databases and the destination database on separate disks to obtain maximum speed.</p> <p>The <tt class="docutils literal"><span class="pre">xapian-compact</span></tt> tool supports an additional option, <tt class="docutils literal"><span class="pre">--multipass</span></tt>, which is useful when merging more than three databases. This will cause the postlist tables to be grouped and merged into temporary tables, which are then grouped and merged, and so on until a single postlist table is created, which is usually faster, but requires more disk space for the temporary files.</p> </div> <div class="section" id="checking-database-integrity"> <h2><a class="toc-backref" href="#id16">Checking database integrity</a></h2> <p>Xapian includes a command-line tool to check that a database is self-consistent. This tool, "xapian-check", runs through the entire database, checking that all the internal nodes are correctly connected. It can also be used on a single table, for example, this command will check the termlist table of database "foo":</p> <pre class="literal-block"> xapian-check foo/termlist.DB </pre> </div> <div class="section" id="converting-a-pre-1-1-4-chert-database-to-a-chert-database"> <h2><a class="toc-backref" href="#id17">Converting a pre-1.1.4 chert database to a chert database</a></h2> <p>The chert format changed in 1.1.4 - at that point the format hadn't been finalised, but a number of users had already deployed it, and it wasn't hard to write an updater, so we provided one called xapian-chert-update which makes a copy with the updated format:</p> <pre class="literal-block"> xapian-chert-update SOURCE DESTINATION </pre> <p>It works much like xapian-compact so should take a similar amount of time (and results in a compact database). The initial version had a few bugs, so use xapian-chert-update from Xapian 1.2.5 or later.</p> <p>The xapian-chert-update utility was removed in Xapian 1.3.0, so you'll need to install Xapian 1.2.x to use it.</p> </div> <div class="section" id="converting-a-flint-database-to-a-chert-database"> <h2><a class="toc-backref" href="#id18">Converting a flint database to a chert database</a></h2> <p>It is possible to convert a flint database to a chert database by using the "copydatabase" example program included with Xapian (you will need to use Xapian 1.2.x for this so it has support for both flint and chert). This is a lot slower to run than "xapian-compact", since it has to perform the sorting of the term occurrence data from scratch, but should be faster than a re-index from source data since it doesn't need to perform the tokenisation step. It is also useful if you no longer have the source data available.</p> <p>The following command will copy a database from "SOURCE" to "DESTINATION", creating the new database at "DESTINATION" as a chert database:</p> <pre class="literal-block"> copydatabase SOURCE DESTINATION </pre> <p>By default copydatabase will renumber your documents starting with docid 1. If the docids are stored in or come from some external system, you should preserve them by using the --no-renumber option (new in Xapian 1.2.5):</p> <pre class="literal-block"> copydatabase --no-renumber SOURCE DESTINATION </pre> </div> <div class="section" id="converting-a-quartz-database-to-a-flint-database"> <h2><a class="toc-backref" href="#id19">Converting a quartz database to a flint database</a></h2> <p>It is possible to convert a quartz database to a flint database by installing Xapian 1.0.x (since this has support for both quartz and flint) and using the "copydatabase" example program included with Xapian. This is a lot slower to run than "xapian-compact", since it has to perform the sorting of the term occurrence data from scratch, but should be faster than a re-index from source data since it doesn't need to perform the tokenisation step. It is also useful if you no longer have the source data available.</p> <p>The following command will copy a database from "SOURCE" to "DESTINATION", creating the new database at "DESTINATION" as a flint database:</p> <pre class="literal-block"> copydatabase SOURCE DESTINATION </pre> </div> <div class="section" id="converting-a-0-9-x-flint-database-to-work-with-1-0-y"> <h2><a class="toc-backref" href="#id20">Converting a 0.9.x flint database to work with 1.0.y</a></h2> <p>In 0.9.x, flint was the development backend.</p> <p>Due to a bug in the flint position list encoding in 0.9.x which made flint databases non-portable between platforms, we had to make an incompatible change in the flint format. It's not easy to write an upgrader, but you can convert a database using the following procedure (although it might be better to rebuild from scratch if you want to use the new UTF-8 support in Xapian::QueryParser, Xapian::Stem, and Xapian::TermGenerator).</p> <p>Run the following command in your Xapian 0.9.x installation to copy your 0.9.x flint database "SOURCE" to a new quartz database "INTERMEDIATE":</p> <pre class="literal-block"> copydatabase SOURCE INTERMEDIATE </pre> <p>Then run the following command in your Xapian 1.0.y installation to copy your quartz database to a 1.0.y flint database "DESTINATION":</p> <pre class="literal-block"> copydatabase INTERMEDIATE DESTINATION </pre> </div> </div> </div> </body> </html>