public-inbox-index(1)

1PUBLIC-INBOX-INDEX(1)      public-inbox user manual      PUBLIC-INBOX-INDEX(1)
2
3
4

NAME

6       public-inbox-index - create and update search indices
7

SYNOPSIS

9       public-inbox-index [OPTIONS] INBOX_DIR...
10
11       public-inbox-index [OPTIONS] --all
12

DESCRIPTION

14       public-inbox-index creates and updates the search, overview and NNTP
15       article number database used by the read-only public-inbox HTTP and
16       NNTP interfaces.  Currently, this requires DBD::SQLite and DBI Perl
17       modules.  Search::Xapian is optional, only to support the PSGI search
18       interface.
19
20       Once the initial indices are created by public-inbox-index,
21       public-inbox-mda(1) and public-inbox-watch(1) will automatically
22       maintain them.
23
24       Running this manually to update indices is only required if relying on
25       git-fetch(1) to mirror an existing public-inbox; or if upgrading to a
26       new version of public-inbox using the "--reindex" option.
27
28       Having the overview and article number database is essential to running
29       the NNTP interface, and strongly recommended for the HTTP interface as
30       it provides thread grouping in addition to normal search functionality.
31

OPTIONS

33       -j JOBS
34       --jobs=JOBS
35           Influences the number of Xapian indexing shards in a
36           (public-inbox-v2-format(5)) inbox.
37
38           See "--jobs" in public-inbox-init(1) for a full description of
39           sharding.
40
41           "--jobs=0" is accepted as of public-inbox 1.6.0 to disable parallel
42           indexing regardless of the number of pre-existing shards.
43
44           If the inbox has not been indexed or initialized, "JOBS - 1" shards
45           will be created (one job is always needed for indexing the overview
46           and article number mapping).
47
48           Default: the number of existing Xapian shards
49
50       -c
51       --compact
52           Compacts the Xapian DBs after indexing.  This is recommended when
53           using "--reindex" to avoid running out of disk space while indexing
54           multiple inboxes.
55
56           While option takes a negligible amount of time compared to
57           "--reindex", it requires temporarily duplicating the entire
58           contents of the Xapian DB.
59
60           This switch may be specified twice, in which case compaction
61           happens both before and after indexing to minimize the temporal
62           footprint of the (re)indexing operation.
63
64           Available since public-inbox 1.4.0.
65
66       --reindex
67           Forces a re-index of all messages in the inbox.  This can be used
68           for in-place upgrades and bugfixes while NNTP/HTTP server processes
69           are utilizing the index.  Keep in mind this roughly doubles the
70           size of the already-large Xapian database.  Using this with
71           "--compact" or running public-inbox-compact(1) afterwards is
72           recommended to release free space.
73
74           public-inbox protects writes to various indices with flock(2), so
75           it is safe to reindex (and rethread) while public-inbox-watch(1),
76           public-inbox-mda(1) or public-inbox-learn(1) run.
77
78           This does not touch the NNTP article number database.  It does not
79           affect threading unless "--rethread" is used.
80
81       --all
82           Index all inboxes configured in ~/.public-inbox/config.  This is an
83           alternative to specifying individual inboxes directories on the
84           command-line.
85
86       --rethread
87           Regenerate internal THREADID and message thread associations when
88           reindexing.
89
90           This fixes some bugs in older versions of public-inbox.  While it
91           is possible to use this without "--reindex", it makes little sense
92           to do so.
93
94           Available in public-inbox 1.6.0+.
95
96       --prune
97           Run git-gc(1) to prune and expire reflogs if discontiguous history
98           is detected.  This is intended to be used in mirrors after running
99           public-inbox-edit(1) or public-inbox-purge(1) to ensure data is
100           expunged from mirrors.
101
102           Available since public-inbox 1.2.0.
103
104       --max-size SIZE
105           Sets or overrides "publicinbox.indexMaxSize" on a per-invocation
106           basis.  See "publicinbox.indexMaxSize" below.
107
108           Available since public-inbox 1.5.0.
109
110       --batch-size SIZE
111           Sets or overrides "publicinbox.indexBatchSize" on a per-invocation
112           basis.  See "publicinbox.indexBatchSize" below.
113
114           When using rotational storage but abundant RAM, using a large value
115           (e.g. "500m") with "--sequential-shard" can significantly speed up
116           and reduce fragmentation during the initial index and full
117           "--reindex" invocations (but not incremental updates).
118
119           Available in public-inbox 1.6.0+.
120
121       --no-fsync
122           Disables fsync(2) and fdatasync(2) operations on SQLite and Xapian.
123           This is only effective with Xapian 1.4+.  This is primarily
124           intended for systems with low RAM and the small (default)
125           "--batch-size=1m".  Users of large "--batch-size" may even find
126           disabling fdatasync(2) causes too much dirty data to accumulate,
127           resulting on latency spikes from writeback.
128
129           Available in public-inbox 1.6.0+.
130
131       --sequential-shard
132           Sets or overrides "publicinbox.indexSequentialShard" on a per-
133           invocation basis.  See "publicinbox.indexSequentialShard" below.
134
135           Available in public-inbox 1.6.0+.
136
137       --skip-docdata
138           Stop storing document data in Xapian on an existing inbox.
139
140           See "--skip-docdata" in public-inbox-init(1) for description and
141           caveats.
142
143           Available in public-inbox 1.6.0+.
144
145       -E EXTINDEX
146       --update-extindex=EXTINDEX
147           Update the given external index (public-inbox-extindex-format(5).
148           Either the configured section name (e.g. "all") or a directory name
149           may be specified.
150
151           Defaults to "all" if "[extindex "all"]" is configured, otherwise no
152           external indices are updated.
153
154           May be specified multiple times in rare cases where multiple
155           external indices are configured.
156
157       --no-update-extindex
158           Do not update the "all" external index by default.  This negates
159           all uses of "-E" / "--update-extindex=" on the command-line.
160
161       --since=DATESTRING
162       --after=DATESTRING
163       --until=DATESTRING
164       --before=DATESTRING
165           Passed directly to git-log(1) to limit changes for "--reindex"
166

FILES

168       For v1 (ssoma) repositories described in public-inbox-v1-format(5).
169       All public-inbox-specific files are contained within the
170       "$GIT_DIR/public-inbox/" directory.
171
172       v2 inboxes are described in public-inbox-v2-format(5).
173

CONFIGURATION

175       publicinbox.indexMaxSize
176               Prevents indexing of messages larger than the specified size
177               value.  A single suffix modifier of "k", "m" or "g" is
178               supported, thus the value of "1m" to prevents indexing of
179               messages larger than one megabyte.
180
181               This is useful for avoiding memory exhaustion in mirrors via
182               git.  It does not prevent public-inbox-mda(1) or
183               public-inbox-watch(1) from importing (and indexing) a message.
184
185               This option is only available in public-inbox 1.5 or later.
186
187               Default: none
188
189       publicinbox.indexBatchSize
190               Flushes changes to the filesystem and releases locks after
191               indexing the given number of bytes.  The default value of "1m"
192               (one megabyte) is low to minimize memory use and reduce
193               contention with parallel invocations of public-inbox-mda(1),
194               public-inbox-learn(1), and public-inbox-watch(1).
195
196               Increase this value on powerful systems to improve throughput
197               at the expense of memory use.  The reduction of lock
198               granularity may not be noticeable on fast systems.  With SSDs,
199               values above "4m" have little benefit.
200
201               For public-inbox-v2-format(5) inboxes, this value is multiplied
202               by the number of Xapian shards.  Thus a typical v2 inbox with 3
203               shards will flush every 3 megabytes by default unless
204               parallelism is disabled via "--sequential-shard" or "--jobs=0".
205
206               This influences memory usage of Xapian, but it is not exact.
207               The actual memory used by Xapian and Perl has been observed in
208               excess of 10x this value.
209
210               This option is available in public-inbox 1.6 or later.  public-
211               inbox 1.5 and earlier used the current default, "1m".
212
213               Default: 1m (one megabyte)
214
215       publicinbox.indexSequentialShard
216               For public-inbox-v2-format(5) inboxes, setting this to "true"
217               allows indexing Xapian shards in multiple passes.  This speeds
218               up indexing on rotational storage with high seek latency by
219               allowing individual shards to fit into the kernel page cache.
220
221               Using a higher-than-normal number of "--jobs" with
222               public-inbox-init(1) may be required to ensure individual
223               shards are small enough to fit into cache.
224
225               Warning: interrupting "public-inbox-index(1)" while this option
226               is in use may leave the search indices out-of-date with respect
227               to SQLite databases.  WWW and IMAP users may notice incomplete
228               search results, but it is otherwise non-fatal.  Using
229               "--reindex" will bring everything back up-to-date.
230
231               Available in public-inbox 1.6.0+.
232
233               This is ignored on public-inbox-v1-format(5) inboxes.
234
235               Default: false, shards are indexed in parallel
236
237       publicinbox.<name>.indexSequentialShard
238               Identical to "publicinbox.indexSequentialShard", but only
239               affect the inbox matching <name>.
240

ENVIRONMENT

242       PI_CONFIG
243               Used to override the default "~/.public-inbox/config" value.
244
245       XAPIAN_FLUSH_THRESHOLD
246               The number of documents to update before committing changes to
247               disk.  This environment is handled directly by Xapian, refer to
248               Xapian API documentation for more details.
249
250               For public-inbox 1.6 and later, use
251               "publicinbox.indexBatchSize" instead.
252
253               Setting "XAPIAN_FLUSH_THRESHOLD" or
254               "publicinbox.indexBatchSize" for a large "--reindex" may cause
255               public-inbox-mda(1), public-inbox-learn(1) and
256               public-inbox-watch(1) tasks to wait long and unpredictable
257               periods of time during "--reindex".
258
259               Default: none, uses "publicinbox.indexBatchSize"
260

UPGRADING

262       Occasionally, public-inbox will update it's schema version and require
263       a full index by running this command.
264

CONTACT

266       Feedback welcome via plain-text mail to <mailto:meta@public-inbox.org>
267
268       The mail archives are hosted at <https://public-inbox.org/meta/> and
269       <http://4uok3hntl7oi7b4uf4rtfwefqeexfzil2w6kgk2jn5z2f764irre7byd.onion/meta/>
270

COPYRIGHT

272       Copyright 2016-2021 all contributors <mailto:meta@public-inbox.org>
273
274       License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
275