1 # Copyright (C) 2020-2021 all contributors <meta@public-inbox.org>
2 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
4 # Detached/external index cross inbox search indexing support
5 # read-write counterpart to PublicInbox::ExtSearch
7 # It's based on the same ideas as public-inbox-v2-format(5) using
8 # over.sqlite3 for dedupe and sharded Xapian. msgmap.sqlite3 is
9 # missing, so there is no Message-ID conflict resolution, meaning
10 # no NNTP support for now.
12 # v2 has a 1:1 mapping of index:inbox or msgmap for NNTP support.
13 # This is intended to be an M:N index:inbox mapping, but it'll likely
14 # be 1:N in common practice (M==1)
16 package PublicInbox::ExtSearchIdx;
19 use parent qw(PublicInbox::ExtSearch PublicInbox::Lock);
20 use Carp qw(croak carp);
21 use Sys::Hostname qw(hostname);
22 use POSIX qw(strftime);
23 use File::Glob qw(bsd_glob GLOB_NOSORT);
24 use PublicInbox::Search;
25 use PublicInbox::SearchIdx qw(prepare_stack is_ancestor is_bad_blob);
26 use PublicInbox::OverIdx;
27 use PublicInbox::MiscIdx;
28 use PublicInbox::MID qw(mids);
29 use PublicInbox::V2Writable;
30 use PublicInbox::InboxWritable;
31 use PublicInbox::ContentHash qw(content_hash);
33 use PublicInbox::DS qw(now add_timer);
34 use DBI qw(:sql_types); # SQL_BLOB
37 my (undef, $dir, $opt) = @_;
38 my $l = $opt->{indexlevel} // 'full';
39 $l !~ $PublicInbox::SearchIdx::INDEXLEVELS and
40 die "invalid indexlevel=$l\n";
41 $l eq 'basic' and die "E: indexlevel=basic not yet supported\n";
43 xpfx => "$dir/ei".PublicInbox::Search::SCHEMA_VERSION,
45 creat => $opt->{creat},
46 ibx_map => {}, # (newsgroup//inboxdir) => $ibx
47 ibx_active => [], # by config section order
48 ibx_known => [], # by config section order
54 lock_path => "$dir/ei.lock",
56 $self->{shards} = $self->count_shards ||
57 nproc_shards({ nproc => $opt->{jobs} });
58 my $oidx = PublicInbox::OverIdx->new("$self->{xpfx}/over.sqlite3");
59 $self->{-no_fsync} = $oidx->{-no_fsync} = 1 if !$opt->{fsync};
60 $self->{oidx} = $oidx;
65 my ($self, $ibx, $types) = @_;
66 $self->{ibx_map}->{$ibx->eidx_key} //= do {
67 delete $self->{-ibx_ary_known}; # invalidate cache
68 delete $self->{-ibx_ary_active}; # invalidate cache
69 $types //= [ qw(active known) ];
71 push @{$self->{"ibx_$t"}}, $ibx;
77 sub _ibx_attach { # each_inbox callback
78 my ($ibx, $self, $types) = @_;
79 attach_inbox($self, $ibx, $types);
83 my ($self, $cfg, $ibxs) = @_;
87 for my $ibx (@$ibxs) {
88 $self->{ibx_map}->{$ibx->eidx_key} //= do {
89 push @{$self->{ibx_active}}, $ibx;
90 push @{$self->{ibx_known}}, $ibx;
94 delete $self->{-ibx_ary_known};
95 delete $self->{-ibx_ary_active};
98 $types //= [ qw(known active) ];
99 $cfg->each_inbox(\&_ibx_attach, $self, $types);
102 sub check_batch_limit ($) {
104 my $self = $req->{self};
105 my $new_smsg = $req->{new_smsg};
106 my $n = $self->{transact_bytes} += $new_smsg->{bytes};
108 # set flag for PublicInbox::V2Writable::index_todo:
109 ${$req->{need_checkpoint}} = 1 if $n >= $self->{batch_bytes};
112 sub apply_boost ($$) {
113 my ($req, $smsg) = @_;
114 my $id2pos = $req->{id2pos}; # index in ibx_sorted
115 my $xr3 = $req->{self}->{oidx}->get_xref3($smsg->{num}, 1);
117 $id2pos->{$a->[0]} <=> $id2pos->{$b->[0]}
119 $a->[1] <=> $b->[1] # break ties with {xnum}
121 my $top_blob = unpack('H*', $xr3->[0]->[2]);
122 my $new_smsg = $req->{new_smsg};
123 return if $top_blob ne $new_smsg->{blob}; # loser
125 # replace the old smsg with the more boosted one
126 $new_smsg->{num} = $smsg->{num};
127 $new_smsg->populate($req->{eml}, $req);
128 $req->{self}->{oidx}->add_overview($req->{eml}, $new_smsg);
132 my ($req, $smsg) = @_;
133 my $self = $req->{self};
134 my $docid = $smsg->{num};
135 my $idx = $self->idx_shard($docid);
136 my $oid = $req->{oid};
137 my $xibx = $req->{ibx};
138 my $eml = $req->{eml};
139 my $eidx_key = $xibx->eidx_key;
140 if (my $new_smsg = $req->{new_smsg}) { # 'm' on cross-posted message
141 my $xnum = $req->{xnum};
142 $self->{oidx}->add_xref3($docid, $xnum, $oid, $eidx_key);
143 $idx->ipc_do('add_eidx_info', $docid, $eidx_key, $eml);
144 apply_boost($req, $smsg) if $req->{boost_in_use};
147 my $nr = $self->{oidx}->remove_xref3($docid, $oid, $eidx_key,
150 $self->{oidx}->eidxq_del($docid);
151 $idx->ipc_do('xdb_remove', $docid);
152 } elsif ($rm_eidx_info) {
153 $idx->ipc_do('remove_eidx_info',
154 $docid, $eidx_key, $eml);
155 $self->{oidx}->eidxq_add($docid); # yes, add
160 # called by V2Writable::sync_prepare
161 sub artnum_max { $_[0]->{oidx}->eidx_max }
163 sub index_unseen ($) {
165 my $new_smsg = $req->{new_smsg} or die 'BUG: {new_smsg} unset';
166 my $eml = delete $req->{eml};
167 $new_smsg->populate($eml, $req);
168 my $self = $req->{self};
169 my $docid = $self->{oidx}->adj_counter('eidx_docid', '+');
170 $new_smsg->{num} = $docid;
171 my $idx = $self->idx_shard($docid);
172 $self->{oidx}->add_overview($eml, $new_smsg);
173 my $oid = $new_smsg->{blob};
174 my $ibx = delete $req->{ibx} or die 'BUG: {ibx} unset';
175 $self->{oidx}->add_xref3($docid, $req->{xnum}, $oid, $ibx->eidx_key);
176 $idx->index_eml($eml, $new_smsg, $ibx->eidx_key);
177 check_batch_limit($req);
180 sub do_finalize ($) {
182 if (my $indexed = $req->{indexed}) { # duplicated messages
183 do_xpost($req, $_) for @$indexed;
184 } elsif (exists $req->{new_smsg}) { # totally unseen messsage
187 # `d' message was already unindexed in the v1/v2 inboxes,
188 # so it's too noisy to warn, here.
190 # cur_cmt may be undef for unindex_oid, set by V2Writable::index_todo
191 if (defined(my $cur_cmt = $req->{cur_cmt})) {
192 ${$req->{latest_cmt}} = $cur_cmt;
196 sub do_step ($) { # main iterator for adding messages to the index
198 my $self = $req->{self} // die 'BUG: {self} missing';
200 if (my $next_arg = $req->{next_arg}) {
201 if (my $smsg = $self->{oidx}->next_by_mid(@$next_arg)) {
202 $req->{cur_smsg} = $smsg;
203 $self->git->cat_async($smsg->{blob},
204 \&ck_existing, $req);
205 return; # ck_existing calls do_step
207 delete $req->{next_arg};
209 die "BUG: {cur_smsg} still set" if $req->{cur_smsg};
210 my $mid = shift(@{$req->{mids}}) // last;
212 $req->{next_arg} = [ $mid, \$id, \$prev ];
218 sub _blob_missing ($$) { # called when $smsg->{blob} is bad
219 my ($req, $smsg) = @_;
220 my $self = $req->{self};
221 my $xref3 = $self->{oidx}->get_xref3($smsg->{num});
222 my @keep = grep(!/:$smsg->{blob}\z/, @$xref3);
224 $keep[0] =~ /:([a-f0-9]{40,}+)\z/ or
225 die "BUG: xref $keep[0] has no OID";
227 $self->{oidx}->remove_xref3($smsg->{num}, $smsg->{blob});
228 my $upd = $self->{oidx}->update_blob($smsg, $oidhex);
229 my $saved = $self->{oidx}->get_art($smsg->{num});
231 $self->{oidx}->delete_by_num($smsg->{num});
235 sub ck_existing { # git->cat_async callback
236 my ($bref, $oid, $type, $size, $req) = @_;
237 my $smsg = delete $req->{cur_smsg} or die 'BUG: {cur_smsg} missing';
238 if ($type eq 'missing') {
239 _blob_missing($req, $smsg);
240 } elsif (!is_bad_blob($oid, $type, $size, $smsg->{blob})) {
241 my $self = $req->{self} // die 'BUG: {self} missing';
242 local $self->{current_info} = "$self->{current_info} $oid";
243 my $cur = PublicInbox::Eml->new($bref);
244 if (content_hash($cur) eq $req->{chash}) {
245 push @{$req->{indexed}}, $smsg; # for do_xpost
246 } # else { index_unseen later }
251 # is the messages visible in the inbox currently being indexed?
252 # return the number if so
253 sub cur_ibx_xnum ($$) {
254 my ($req, $bref) = @_;
255 my $ibx = $req->{ibx} or die 'BUG: current {ibx} missing';
257 $req->{eml} = PublicInbox::Eml->new($bref);
258 $req->{chash} = content_hash($req->{eml});
259 $req->{mids} = mids($req->{eml});
260 for my $mid (@{$req->{mids}}) {
262 while (my $x = $ibx->over->next_by_mid($mid, \$id, \$prev)) {
263 return $x->{num} if $x->{blob} eq $req->{oid};
269 sub index_oid { # git->cat_async callback for 'm'
270 my ($bref, $oid, $type, $size, $req) = @_;
271 my $self = $req->{self};
272 local $self->{current_info} = "$self->{current_info} $oid";
273 return if is_bad_blob($oid, $type, $size, $req->{oid});
274 my $new_smsg = $req->{new_smsg} = bless {
276 }, 'PublicInbox::Smsg';
277 $new_smsg->set_bytes($$bref, $size);
278 defined($req->{xnum} = cur_ibx_xnum($req, $bref)) or return;
283 sub unindex_oid { # git->cat_async callback for 'd'
284 my ($bref, $oid, $type, $size, $req) = @_;
285 my $self = $req->{self};
286 local $self->{current_info} = "$self->{current_info} $oid";
287 return if is_bad_blob($oid, $type, $size, $req->{oid});
288 return if defined(cur_ibx_xnum($req, $bref)); # was re-added
292 # overrides V2Writable::last_commits, called by sync_ranges via sync_prepare
294 my ($self, $sync) = @_;
296 my $ekey = $sync->{ibx}->eidx_key;
297 my $uv = $sync->{ibx}->uidvalidity;
298 for my $i (0..$sync->{epoch_max}) {
299 $heads->[$i] = $self->{oidx}->eidx_meta("lc-v2:$ekey//$uv;$i");
304 sub _ibx_index_reject ($) {
306 $ibx->mm // return 'unindexed, no msgmap.sqlite3';
307 $ibx->uidvalidity // return 'no UIDVALIDITY';
308 $ibx->over // return 'unindexed, no over.sqlite3';
312 sub _sync_inbox ($$$) {
313 my ($self, $sync, $ibx) = @_;
314 my $ekey = $ibx->eidx_key;
315 if (defined(my $err = _ibx_index_reject($ibx))) {
316 return "W: skipping $ekey ($err)";
319 $sync->{nr} = \(my $nr = 0);
320 my $v = $ibx->version;
322 $sync->{epoch_max} = $ibx->max_git_epoch // return;
323 sync_prepare($self, $sync); # or return # TODO: once MiscIdx is stable
325 my $uv = $ibx->uidvalidity;
326 my $lc = $self->{oidx}->eidx_meta("lc-v1:$ekey//$uv");
327 my $head = $ibx->mm->last_commit //
328 return "E: $ibx->{inboxdir} is not indexed";
329 my $stk = prepare_stack($sync, $lc ? "$lc..$head" : $head);
330 my $unit = { stack => $stk, git => $ibx->git };
331 push @{$sync->{todo}}, $unit;
333 return "E: $ekey unsupported inbox version (v$v)";
335 for my $unit (@{delete($sync->{todo}) // []}) {
336 last if $sync->{quit};
337 index_todo($self, $sync, $unit);
339 $self->{midx}->index_ibx($ibx) unless $sync->{quit};
340 $ibx->git->cleanup; # done with this inbox, now
344 sub gc_unref_doc ($$$$) {
345 my ($self, $ibx_id, $eidx_key, $docid) = @_;
347 # for debug/info purposes, oids may no longer be accessible
348 my $dbh = $self->{oidx}->dbh;
349 my $sth = $dbh->prepare_cached(<<'', undef, 1);
350 SELECT oidbin FROM xref3 WHERE docid = ? AND ibx_id = ?
352 $sth->execute($docid, $ibx_id);
353 my @oid = map { unpack('H*', $_->[0]) } @{$sth->fetchall_arrayref};
355 $remain += $self->{oidx}->remove_xref3($docid, $oid, $eidx_key);
358 $self->{oidx}->eidxq_add($docid); # enqueue for reindex
360 warn "I: unref #$docid $eidx_key $oid\n";
363 warn "I: remove #$docid $eidx_key @oid\n";
364 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
369 my ($self, $opt) = @_;
370 $self->{cfg} or die "E: GC requires ->attach_config\n";
372 $self->idx_init($opt); # acquire lock via V2Writable::_idx_init
374 my $dbh = $self->{oidx}->dbh;
375 $dbh->do('PRAGMA case_sensitive_like = ON'); # only place we use LIKE
376 my $x3_doc = $dbh->prepare('SELECT docid FROM xref3 WHERE ibx_id = ?');
377 my $ibx_ck = $dbh->prepare('SELECT ibx_id,eidx_key FROM inboxes');
378 my $lc_i = $dbh->prepare(<<'');
379 SELECT key FROM eidx_meta WHERE key LIKE ? ESCAPE ?
382 while (my ($ibx_id, $eidx_key) = $ibx_ck->fetchrow_array) {
383 next if $self->{ibx_map}->{$eidx_key};
384 $self->{midx}->remove_eidx_key($eidx_key);
385 warn "I: deleting messages for $eidx_key...\n";
386 $x3_doc->execute($ibx_id);
387 while (defined(my $docid = $x3_doc->fetchrow_array)) {
388 gc_unref_doc($self, $ibx_id, $eidx_key, $docid);
390 $dbh->prepare_cached(<<'')->execute($ibx_id);
391 DELETE FROM inboxes WHERE ibx_id = ?
393 # drop last_commit info
395 $pat =~ s/([_%\\])/\\$1/g;
396 $lc_i->execute("lc-%:$pat//%", '\\');
397 while (my ($key) = $lc_i->fetchrow_array) {
398 next if $key !~ m!\Alc-v[1-9]+:\Q$eidx_key\E//!;
399 warn "I: removing $key\n";
400 $dbh->prepare_cached(<<'')->execute($key);
401 DELETE FROM eidx_meta WHERE key = ?
405 warn "I: $eidx_key removed\n";
408 # it's not real unless it's in `over', we use parallelism here,
409 # shards will be reading directly from over, so commit
410 $self->{oidx}->commit_lazy;
411 $self->{oidx}->begin_lazy;
413 for my $idx (@{$self->{idx_shards}}) {
414 warn "I: cleaning up shard #$idx->{shard}\n";
415 $idx->shard_over_check($self->{oidx});
417 my $nr = $dbh->do(<<'');
418 DELETE FROM xref3 WHERE docid NOT IN (SELECT num FROM over)
420 warn "I: eliminated $nr stale xref3 entries\n" if $nr != 0;
422 # fixup from old bugs:
423 $nr = $dbh->do(<<'');
424 DELETE FROM over WHERE num NOT IN (SELECT docid FROM xref3)
426 warn "I: eliminated $nr stale over entries\n" if $nr != 0;
431 my ($self, $sync, $smsg) = @_;
432 my $ibx_id = delete($smsg->{ibx_id}) // die '{ibx_id} unset';
433 my $pos = $sync->{id2pos}->{$ibx_id} // die "$ibx_id no pos";
434 $self->{-ibx_ary_known}->[$pos] //
435 die "BUG: ibx for $smsg->{blob} not mapped"
438 sub _fd_constrained ($) {
440 $self->{-fd_constrained} //= do {
442 if (eval { require BSD::Resource; 1 }) {
443 my $NOFILE = BSD::Resource::RLIMIT_NOFILE();
444 ($soft, undef) = BSD::Resource::getrlimit($NOFILE);
446 chomp($soft = `sh -c 'ulimit -n'`);
448 if (defined($soft)) {
449 # $want is an estimate
450 my $want = scalar(@{$self->{ibx_active}}) + 64;
451 my $ret = $want > $soft;
454 RLIMIT_NOFILE=$soft insufficient (want: $want), will close DB handles early
459 warn "Unable to determine RLIMIT_NOFILE: $@\n";
465 sub _reindex_finalize ($$$) {
466 my ($req, $smsg, $eml) = @_;
467 my $sync = $req->{sync};
468 my $self = $sync->{self};
469 my $by_chash = delete $req->{by_chash} or die 'BUG: no {by_chash}';
470 my $nr = scalar(keys(%$by_chash)) or die 'BUG: no content hashes';
471 my $orig_smsg = $req->{orig_smsg} // die 'BUG: no {orig_smsg}';
472 my $docid = $smsg->{num} = $orig_smsg->{num};
473 $self->{oidx}->add_overview($eml, $smsg); # may rethread
474 check_batch_limit({ %$sync, new_smsg => $smsg });
475 my $chash0 = $smsg->{chash} // die "BUG: $smsg->{blob} no {chash}";
476 my $stable = delete($by_chash->{$chash0}) //
477 die "BUG: $smsg->{blob} chash missing";
478 my $idx = $self->idx_shard($docid);
479 my $top_smsg = pop @$stable;
480 $top_smsg == $smsg or die 'BUG: top_smsg != smsg';
481 my $ibx = _ibx_for($self, $sync, $smsg);
482 $idx->index_eml($eml, $smsg, $ibx->eidx_key);
483 for my $x (reverse @$stable) {
484 $ibx = _ibx_for($self, $sync, $x);
485 my $hdr = delete $x->{hdr} // die 'BUG: no {hdr}';
486 $idx->ipc_do('add_eidx_info', $docid, $ibx->eidx_key, $hdr);
488 return if $nr == 1; # likely, all good
490 warn "W: #$docid split into $nr due to deduplication change\n";
492 for my $ary (values %$by_chash) {
493 for my $x (reverse @$ary) {
494 warn "removing #$docid xref3 $x->{blob}\n";
495 my $n = $self->{oidx}->remove_xref3($docid, $x->{blob});
496 die "BUG: $x->{blob} invalidated #$docid" if $n == 0;
498 my $x = pop(@$ary) // die "BUG: #$docid {by_chash} empty";
499 $x->{num} = delete($x->{xnum}) // die '{xnum} unset';
500 $ibx = _ibx_for($self, $sync, $x);
501 if (my $over = $ibx->over) {
502 my $e = $over->get_art($x->{num});
503 $e->{blob} eq $x->{blob} or die <<EOF;
504 $x->{blob} != $e->{blob} (${\$ibx->eidx_key}:$e->{num});
506 push @todo, $ibx, $e;
507 $over->dbh_close if _fd_constrained($self);
509 die "$ibx->{inboxdir}: over.sqlite3 unusable: $!\n";
513 while (my ($ibx, $e) = splice(@todo, 0, 2)) {
514 reindex_unseen($self, $sync, $ibx, $e);
518 sub _reindex_oid { # git->cat_async callback
519 my ($bref, $oid, $type, $size, $req) = @_;
520 my $sync = $req->{sync};
521 my $self = $sync->{self};
522 my $orig_smsg = $req->{orig_smsg} // die 'BUG: no {orig_smsg}';
523 my $expect_oid = $req->{xr3r}->[$req->{ix}]->[2];
524 my $docid = $orig_smsg->{num};
525 if (is_bad_blob($oid, $type, $size, $expect_oid)) {
526 my $remain = $self->{oidx}->remove_xref3($docid, $expect_oid);
528 warn "W: #$docid gone or corrupted\n";
529 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
530 } elsif (my $next_oid = $req->{xr3r}->[++$req->{ix}]->[2]) {
531 $self->git->cat_async($next_oid, \&_reindex_oid, $req);
533 warn "BUG: #$docid gone (UNEXPECTED)\n";
534 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
538 my $ci = $self->{current_info};
539 local $self->{current_info} = "$ci #$docid $oid";
540 my $re_smsg = bless { blob => $oid }, 'PublicInbox::Smsg';
541 $re_smsg->set_bytes($$bref, $size);
542 my $eml = PublicInbox::Eml->new($bref);
543 $re_smsg->populate($eml, { autime => $orig_smsg->{ds},
544 cotime => $orig_smsg->{ts} });
545 my $chash = content_hash($eml);
546 $re_smsg->{chash} = $chash;
547 $re_smsg->{xnum} = $req->{xr3r}->[$req->{ix}]->[1];
548 $re_smsg->{ibx_id} = $req->{xr3r}->[$req->{ix}]->[0];
549 $re_smsg->{hdr} = $eml->header_obj;
550 push @{$req->{by_chash}->{$chash}}, $re_smsg;
551 if (my $next_oid = $req->{xr3r}->[++$req->{ix}]->[2]) {
552 $self->git->cat_async($next_oid, \&_reindex_oid, $req);
553 } else { # last $re_smsg is the highest priority xref3
554 local $self->{current_info} = "$ci #$docid";
555 _reindex_finalize($req, $re_smsg, $eml);
559 sub _reindex_smsg ($$$) {
560 my ($self, $sync, $smsg) = @_;
561 my $docid = $smsg->{num};
562 my $xr3 = $self->{oidx}->get_xref3($docid, 1);
563 if (scalar(@$xr3) == 0) { # _reindex_check_stale should've covered this
565 BUG? #$docid $smsg->{blob} is not referenced by inboxes during reindex
567 $self->{oidx}->delete_by_num($docid);
568 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
572 # we sort {xr3r} in the reverse order of ibx_sorted so we can
573 # hit the common case in _reindex_finalize without rereading
574 # from git (or holding multiple messages in memory).
575 my $id2pos = $sync->{id2pos}; # index in ibx_sorted
577 $id2pos->{$b->[0]} <=> $id2pos->{$a->[0]}
579 $b->[1] <=> $a->[1] # break ties with {xnum}
581 @$xr3 = map { [ $_->[0], $_->[1], unpack('H*', $_->[2]) ] } @$xr3;
582 my $req = { orig_smsg => $smsg, sync => $sync, xr3r => $xr3, ix => 0 };
583 $self->git->cat_async($xr3->[$req->{ix}]->[2], \&_reindex_oid, $req);
586 sub checkpoint_due ($) {
588 ${$sync->{need_checkpoint}} || (now() > $sync->{next_check});
592 # I've copied FS images and only changed the hostname before,
593 # so prepend hostname. Use `state' since these a BOFH can change
594 # these while this process is running and we always want to be
595 # able to release locks taken by this process.
596 state $retval = hostname . '-' . do {
597 my $m; # machine-id(5) is systemd
598 if (open(my $fh, '<', '/etc/machine-id')) { $m = <$fh> }
599 # (g)hostid(1) is in GNU coreutils, kern.hostid is most BSDs
600 chomp($m ||= `{ sysctl -n kern.hostid ||
601 hostid || ghostid; } 2>/dev/null`
602 || "no-machine-id-or-hostid-on-$^O");
609 my $expect = delete($self->{-eidxq_locked}) or return;
610 my ($owner_pid, undef) = split(/-/, $expect);
611 return if $owner_pid != $$; # shards may fork
612 my $oidx = $self->{oidx};
614 my $cur = $oidx->eidx_meta('eidxq_lock') // '';
615 if ($cur eq $expect) {
616 $oidx->eidx_meta('eidxq_lock', '');
618 } elsif ($cur ne '') {
619 warn "E: eidxq_lock($expect) stolen by $cur\n";
621 warn "E: eidxq_lock($expect) released by another process\n";
628 eidxq_release($self) and $self->{oidx}->commit_lazy;
631 sub _eidxq_take ($) {
633 my $val = "$$-${\time}-$>-".host_ident;
634 $self->{oidx}->eidx_meta('eidxq_lock', $val);
635 $self->{-eidxq_locked} = $val;
638 sub eidxq_lock_acquire ($) {
640 my $oidx = $self->{oidx};
642 my $cur = $oidx->eidx_meta('eidxq_lock') || return _eidxq_take($self);
643 if (my $locked = $self->{-eidxq_locked}) { # be lazy
644 return $locked if $locked eq $cur;
646 my ($pid, $time, $euid, $ident) = split(/-/, $cur, 4);
647 my $t = strftime('%Y-%m-%d %k:%M:%S', gmtime($time));
648 if ($euid == $> && $ident eq host_ident) {
651 I: PID:$pid (re)indexing Xapian since $t, it will continue our work
655 warn "I: eidxq_lock is stale ($cur), clobbering\n";
656 return _eidxq_take($self);
658 warn "E: kill(0, $pid) failed: $!\n"; # fall-through:
660 my $fn = $oidx->dbh->sqlite_db_filename;
662 W: PID:$pid, UID:$euid on $ident is indexing Xapian since $t
663 W: If this is unexpected, delete `eidxq_lock' from the `eidx_meta' table:
664 W: sqlite3 $fn 'DELETE FROM eidx_meta WHERE key = "eidxq_lock"'
669 sub ibx_sorted ($$) {
670 my ($self, $type) = @_;
671 $self->{"-ibx_ary_$type"} //= do {
672 # highest boost first, stable for config-ordering tiebreaker
675 ($b->{boost} // 0) <=> ($a->{boost} // 0)
676 } @{$self->{'ibx_'.$type} // die "BUG: $type unknown"} ];
680 sub prep_id2pos ($) {
684 $id2pos{$_->{-ibx_id}} = $pos++ for (@{ibx_sorted($self, 'known')});
688 sub eidxq_process ($$) { # for reindexing
689 my ($self, $sync) = @_;
691 return unless eidxq_lock_acquire($self);
692 my $dbh = $self->{oidx}->dbh;
693 my $tot = $dbh->selectrow_array('SELECT COUNT(*) FROM eidxq') or return;
695 local $sync->{-regen_fmt} = "%u/$tot\n";
696 my $pr = $sync->{-opt}->{-progress};
698 my $min = $dbh->selectrow_array('SELECT MIN(docid) FROM eidxq');
699 my $max = $dbh->selectrow_array('SELECT MAX(docid) FROM eidxq');
700 $pr->("Xapian indexing $min..$max (total=$tot)\n");
702 $sync->{id2pos} //= prep_id2pos($self);
705 $del = $dbh->prepare('DELETE FROM eidxq WHERE docid = ?');
706 $iter = $dbh->prepare('SELECT docid FROM eidxq ORDER BY docid ASC');
708 while (defined(my $docid = $iter->fetchrow_array)) {
709 last if $sync->{quit};
710 if (my $smsg = $self->{oidx}->get_art($docid)) {
711 _reindex_smsg($self, $sync, $smsg);
713 warn "E: #$docid does not exist in over\n";
715 $del->execute($docid);
718 if (checkpoint_due($sync)) {
719 $dbh = $del = $iter = undef;
720 reindex_checkpoint($self, $sync); # release lock
721 $dbh = $self->{oidx}->dbh;
725 $self->git->async_wait_all;
726 $pr->("reindexed ${$sync->{nr}}/$tot\n") if $pr;
729 sub _reindex_unseen { # git->cat_async callback
730 my ($bref, $oid, $type, $size, $req) = @_;
731 return if is_bad_blob($oid, $type, $size, $req->{oid});
732 my $self = $req->{self} // die 'BUG: {self} unset';
733 local $self->{current_info} = "$self->{current_info} $oid";
734 my $new_smsg = bless { blob => $oid, }, 'PublicInbox::Smsg';
735 $new_smsg->set_bytes($$bref, $size);
736 my $eml = $req->{eml} = PublicInbox::Eml->new($bref);
737 $req->{new_smsg} = $new_smsg;
738 $req->{chash} = content_hash($eml);
739 $req->{mids} = mids($eml); # do_step iterates through this
740 do_step($req); # enter the normal indexing flow
743 # --reindex may catch totally unseen messages, this handles them
744 sub reindex_unseen ($$$$) {
745 my ($self, $sync, $ibx, $xsmsg) = @_;
748 autime => $xsmsg->{ds},
749 cotime => $xsmsg->{ts},
750 oid => $xsmsg->{blob},
752 xnum => $xsmsg->{num},
753 # {mids} and {chash} will be filled in at _reindex_unseen
755 warn "I: reindex_unseen ${\$ibx->eidx_key}:$req->{xnum}:$req->{oid}\n";
756 $self->git->cat_async($xsmsg->{blob}, \&_reindex_unseen, $req);
759 sub _reindex_check_unseen ($$$) {
760 my ($self, $sync, $ibx) = @_;
761 my $ibx_id = $ibx->{-ibx_id};
763 my ($beg, $end) = (1, $slice);
765 # first, check if we missed any messages in target $ibx
767 my $pr = $sync->{-opt}->{-progress};
768 my $ekey = $ibx->eidx_key;
769 local $sync->{-regen_fmt} =
770 "$ekey checking unseen %u/".$ibx->over->max."\n";
773 while (scalar(@{$msgs = $ibx->over->query_xover($beg, $end)})) {
774 ${$sync->{nr}} = $beg;
775 $beg = $msgs->[-1]->{num} + 1;
776 $end = $beg + $slice;
777 if (checkpoint_due($sync)) {
778 reindex_checkpoint($self, $sync); # release lock
781 my $inx3 = $self->{oidx}->dbh->prepare_cached(<<'', undef, 1);
782 SELECT DISTINCT(docid) FROM xref3 WHERE
783 ibx_id = ? AND xnum = ? AND oidbin = ?
785 for my $xsmsg (@$msgs) {
786 my $oidbin = pack('H*', $xsmsg->{blob});
787 $inx3->bind_param(1, $ibx_id);
788 $inx3->bind_param(2, $xsmsg->{num});
789 $inx3->bind_param(3, $oidbin, SQL_BLOB);
791 my $docids = $inx3->fetchall_arrayref;
792 # index messages which were totally missed
793 # the first time around ASAP:
794 if (scalar(@$docids) == 0) {
795 reindex_unseen($self, $sync, $ibx, $xsmsg);
796 } else { # already seen, reindex later
797 for my $r (@$docids) {
798 $self->{oidx}->eidxq_add($r->[0]);
801 last if $sync->{quit};
803 last if $sync->{quit};
807 sub _reindex_check_stale ($$$) {
808 my ($self, $sync, $ibx) = @_;
810 my $pr = $sync->{-opt}->{-progress};
812 my $ekey = $ibx->eidx_key;
813 local $sync->{-regen_fmt} =
814 "$ekey check stale/missing %u/".$ibx->over->max."\n";
817 if (checkpoint_due($sync)) {
818 reindex_checkpoint($self, $sync); # release lock
820 # now, check if there's stale xrefs
821 my $iter = $self->{oidx}->dbh->prepare_cached(<<'', undef, 1);
822 SELECT docid,xnum,oidbin FROM xref3 WHERE ibx_id = ? AND docid > ?
823 ORDER BY docid,xnum ASC LIMIT 10000
825 $iter->execute($ibx->{-ibx_id}, $min);
828 while (my ($docid, $xnum, $oidbin) = $iter->fetchrow_array) {
829 return if $sync->{quit};
830 ${$sync->{nr}} = $xnum;
832 $fetching = $min = $docid;
833 my $smsg = $ibx->over->get_art($xnum);
837 } elsif (pack('H*', $smsg->{blob}) ne $oidbin) {
838 $err = "mismatch (!= $smsg->{blob})";
840 next; # likely, all good
842 # current_info already has eidx_key
843 my $oidhex = unpack('H*', $oidbin);
844 warn "$xnum:$oidhex (#$docid): $err\n";
845 my $del = $self->{oidx}->dbh->prepare_cached(<<'');
846 DELETE FROM xref3 WHERE ibx_id = ? AND xnum = ? AND oidbin = ?
848 $del->bind_param(1, $ibx->{-ibx_id});
849 $del->bind_param(2, $xnum);
850 $del->bind_param(3, $oidbin, SQL_BLOB);
853 # get_xref3 over-fetches, but this is a rare path:
854 my $xr3 = $self->{oidx}->get_xref3($docid);
855 my $idx = $self->idx_shard($docid);
856 if (scalar(@$xr3) == 0) { # all gone
857 $self->{oidx}->delete_by_num($docid);
858 $self->{oidx}->eidxq_del($docid);
859 $idx->ipc_do('xdb_remove', $docid);
860 } else { # enqueue for reindex of remaining messages
861 $idx->ipc_do('remove_eidx_info',
862 $docid, $ibx->eidx_key);
863 $self->{oidx}->eidxq_add($docid); # yes, add
866 } while (defined $fetching);
869 sub _reindex_inbox ($$$) {
870 my ($self, $sync, $ibx) = @_;
871 my $ekey = $ibx->eidx_key;
872 local $self->{current_info} = $ekey;
873 if (defined(my $err = _ibx_index_reject($ibx))) {
874 warn "W: cannot reindex $ekey ($err)\n";
876 _reindex_check_unseen($self, $sync, $ibx);
877 _reindex_check_stale($self, $sync, $ibx) unless $sync->{quit};
879 delete @$ibx{qw(over mm search git)}; # won't need these for a bit
883 my ($self, $sync) = @_;
885 # acquire eidxq_lock early because full reindex takes forever
886 # and incremental -extindex processes can run during our checkpoints
887 if (!eidxq_lock_acquire($self)) {
888 warn "E: aborting --reindex\n";
891 for my $ibx (@{ibx_sorted($self, 'active')}) {
892 _reindex_inbox($self, $sync, $ibx);
893 last if $sync->{quit};
895 $self->git->async_wait_all; # ensure eidxq gets filled completely
896 eidxq_process($self, $sync) unless $sync->{quit};
900 my ($self, $sync, $ibx) = @_;
901 my $err = _sync_inbox($self, $sync, $ibx);
902 delete @$ibx{qw(mm over)};
903 warn $err, "\n" if defined($err);
906 sub dd_smsg { # git->cat_async callback
907 my ($bref, $oid, $type, $size, $dd) = @_;
908 my $smsg = $dd->{smsg} // die 'BUG: dd->{smsg} missing';
909 my $self = $dd->{self} // die 'BUG: {self} missing';
910 my $per_mid = $dd->{per_mid} // die 'BUG: {per_mid} missing';
911 if ($type eq 'missing') {
912 _blob_missing($dd, $smsg);
913 } elsif (!is_bad_blob($oid, $type, $size, $smsg->{blob})) {
914 local $self->{current_info} = "$self->{current_info} $oid";
915 my $chash = content_hash(PublicInbox::Eml->new($bref));
916 push(@{$per_mid->{dd_chash}->{$chash}}, $smsg);
918 return if $per_mid->{last_smsg} != $smsg;
919 while (my ($chash, $ary) = each %{$per_mid->{dd_chash}}) {
920 my $keep = shift @$ary;
921 next if !scalar(@$ary);
922 $per_mid->{sync}->{dedupe_cull} += scalar(@$ary);
924 "# <$keep->{mid}> keeping #$keep->{num}, dropping ",
925 join(', ', map { "#$_->{num}" } @$ary),"\n";
926 next if $per_mid->{sync}->{-opt}->{'dry-run'};
927 my $oidx = $self->{oidx};
928 for my $smsg (@$ary) {
929 my $gone = $smsg->{num};
930 $oidx->merge_xref3($keep->{num}, $gone, $smsg->{blob});
931 $self->idx_shard($gone)->ipc_do('xdb_remove', $gone);
932 $oidx->delete_by_num($gone);
937 sub eidx_dedupe ($$$) {
938 my ($self, $sync, $msgids) = @_;
939 $sync->{dedupe_cull} = 0;
942 return unless eidxq_lock_acquire($self);
943 my ($iter, $cur_mid);
946 my ($max_id) = $self->{oidx}->dbh->selectrow_array(<<EOS);
947 SELECT MAX(id) FROM msgid
949 local $sync->{-regen_fmt} = "dedupe %u/$max_id\n";
951 # note: we could write this query more intelligently,
952 # but that causes lock contention with read-only processes
954 $cur_mid = $msgids->[$idx];
955 if ($cur_mid eq '') { # all Message-IDs
956 $iter = $self->{oidx}->dbh->prepare(<<EOS);
957 SELECT mid,id FROM msgid WHERE id > ? ORDER BY id ASC
959 $iter->execute($min_id);
961 $iter = $self->{oidx}->dbh->prepare(<<EOS);
962 SELECT mid,id FROM msgid WHERE mid = ? AND id > ? ORDER BY id ASC
964 $iter->execute($cur_mid, $min_id);
966 while (my ($mid, $id) = $iter->fetchrow_array) {
967 last if $sync->{quit};
968 $self->{current_info} = "dedupe $mid";
969 ${$sync->{nr}} = $min_id = $id;
971 while (my $x = $self->{oidx}->next_by_mid($mid, \$id, \$prv)) {
974 next if scalar(@smsg) < 2;
976 dd_chash => {}, # chash => [ary of smsgs]
977 last_smsg => $smsg[-1],
981 $candidates += scalar(@smsg) - 1;
982 for my $smsg (@smsg) {
988 $self->git->cat_async($smsg->{blob}, \&dd_smsg, $dd);
990 # need to wait on every single one @smsg contents can get
991 # invalidated inside dd_smsg for messages with multiple
993 $self->git->async_wait_all;
995 if (checkpoint_due($sync)) {
997 reindex_checkpoint($self, $sync);
1001 goto dedupe_restart if defined($msgids->[++$idx]);
1003 my $n = delete $sync->{dedupe_cull};
1004 if (my $pr = $sync->{-opt}->{-progress}) {
1005 $pr->("culled $n/$candidates candidates ($nr_mid msgids)\n");
1010 sub eidx_sync { # main entry point
1011 my ($self, $opt) = @_;
1013 my $warn_cb = $SIG{__WARN__} || \&CORE::warn;
1014 local $self->{current_info} = '';
1015 local $SIG{__WARN__} = sub {
1016 return if PublicInbox::Eml::warn_ignore(@_);
1017 $warn_cb->($self->{current_info}, ': ', @_);
1019 $self->idx_init($opt); # acquire lock via V2Writable::_idx_init
1020 $self->{oidx}->rethread_prepare($opt);
1022 need_checkpoint => \(my $need_checkpoint = 0),
1024 next_check => now() + 10,
1026 # DO NOT SET {reindex} here, it's incompatible with reused
1027 # V2Writable code, reindex is totally different here
1028 # compared to v1/v2 inboxes because we have multiple histories
1030 -regen_fmt => "%u/?\n",
1032 local $SIG{USR1} = sub { $need_checkpoint = 1 };
1033 my $quit = PublicInbox::SearchIdx::quit_cb($sync);
1034 local $SIG{QUIT} = $quit;
1035 local $SIG{INT} = $quit;
1036 local $SIG{TERM} = $quit;
1037 for my $ibx (@{ibx_sorted($self, 'known')}) {
1038 $ibx->{-ibx_id} //= $self->{oidx}->ibx_id($ibx->eidx_key);
1041 if (scalar(grep { defined($_->{boost}) } @{$self->{ibx_known}})) {
1042 $sync->{id2pos} //= prep_id2pos($self);
1043 $sync->{boost_in_use} = 1;
1046 if (my $msgids = delete($opt->{dedupe})) {
1047 local $sync->{checkpoint_unlocks} = 1;
1048 eidx_dedupe($self, $sync, $msgids);
1050 if (delete($opt->{reindex})) {
1051 local $sync->{checkpoint_unlocks} = 1;
1052 eidx_reindex($self, $sync);
1055 # don't use $_ here, it'll get clobbered by reindex_checkpoint
1056 if ($opt->{scan} // 1) {
1057 for my $ibx (@{ibx_sorted($self, 'active')}) {
1058 last if $sync->{quit};
1059 sync_inbox($self, $sync, $ibx);
1062 $self->{oidx}->rethread_done($opt) unless $sync->{quit};
1063 eidxq_process($self, $sync) unless $sync->{quit};
1065 eidxq_release($self);
1067 $sync; # for eidx_watch
1070 sub update_last_commit { # overrides V2Writable
1071 my ($self, $sync, $stk) = @_;
1072 my $unit = $sync->{unit} // return;
1073 my $latest_cmt = $stk ? $stk->{latest_cmt} : ${$sync->{latest_cmt}};
1074 defined($latest_cmt) or return;
1075 my $ibx = $sync->{ibx} or die 'BUG: {ibx} missing';
1076 my $ekey = $ibx->eidx_key;
1077 my $uv = $ibx->uidvalidity;
1078 my $epoch = $unit->{epoch};
1080 my $v = $ibx->version;
1082 die 'No {epoch} for v2 unit' unless defined $epoch;
1083 $meta_key = "lc-v2:$ekey//$uv;$epoch";
1085 die 'Unexpected {epoch} for v1 unit' if defined $epoch;
1086 $meta_key = "lc-v1:$ekey//$uv";
1088 die "Unsupported inbox version: $v";
1090 my $last = $self->{oidx}->eidx_meta($meta_key);
1091 if (defined $last && is_ancestor($self->git, $last, $latest_cmt)) {
1092 my @cmd = (qw(rev-list --count), "$last..$latest_cmt");
1093 chomp(my $n = $unit->{git}->qx(@cmd));
1094 return if $n ne '' && $n == 0;
1096 $self->{oidx}->eidx_meta($meta_key, $latest_cmt);
1099 sub _idx_init { # with_umask callback
1100 my ($self, $opt) = @_;
1101 PublicInbox::V2Writable::_idx_init($self, $opt); # acquires ei.lock
1102 $self->{midx} = PublicInbox::MiscIdx->new($self);
1105 sub symlink_packs ($$) {
1106 my ($ibx, $pd) = @_;
1108 my $glob = "$ibx->{inboxdir}/git/*.git/objects/pack/*.idx";
1109 for my $idx (bsd_glob($glob, GLOB_NOSORT)) {
1110 my $src = substr($idx, 0, -length('.idx'));
1111 my $dst = $pd . substr($src, rindex($src, '/'));
1112 if (-f "$src.pack" and
1113 symlink("$src.pack", "$dst.pack") and
1114 symlink($idx, "$dst.idx") and
1117 # .promisor, .bitmap, .rev and .keep are optional
1118 # XXX should we symlink .keep here?
1119 for my $s (qw(promisor bitmap rev)) {
1120 symlink("$src.$s", "$dst.$s") if -f "$src.$s";
1122 } elsif (!$!{EEXIST}) {
1123 warn "W: ln -s $src.{pack,idx} => $dst.*: $!\n";
1124 unlink "$dst.pack", "$dst.idx";
1130 sub idx_init { # similar to V2Writable
1131 my ($self, $opt) = @_;
1132 return if $self->{idx_shards};
1134 $self->git->cleanup;
1136 my $ALL = $self->git->{git_dir}; # ALL.git
1138 if ($opt->{-private}) { # LeiStore
1141 umask 077; # don't bother restoring
1142 PublicInbox::Import::init_bare($ALL);
1143 $self->git->qx(qw(config core.sharedRepository 0600));
1146 PublicInbox::Import::init_bare($ALL) unless $old;
1148 my $info_dir = "$ALL/objects/info";
1149 my $alt = "$info_dir/alternates";
1150 my (@old, @new, %seen); # seen: st_dev + st_ino
1152 open(my $fh, '<', $alt) or die "open $alt: $!";
1153 $mode = (stat($fh))[2] & 07777;
1154 while (my $line = <$fh>) {
1155 chomp(my $d = $line);
1157 # expand relative path (/local/ stuff)
1158 substr($d, 0, 3) eq '../' and
1159 $d = "$ALL/objects/$d";
1160 if (my @st = stat($d)) {
1161 next if $seen{"$st[0]\0$st[1]"}++;
1163 warn "W: stat($d) failed (from $alt): $!\n";
1164 next if $opt->{-idx_gc};
1170 # for LeiStore, and possibly some mirror-only state
1171 if (opendir(my $dh, my $local = "$self->{topdir}/local")) {
1172 # highest numbered epoch first
1173 for my $n (sort { $b <=> $a } map { substr($_, 0, -4) + 0 }
1174 grep(/\A[0-9]+\.git\z/, readdir($dh))) {
1175 my $d = "$local/$n.git/objects"; # absolute path
1176 if (my @st = stat($d)) {
1177 next if $seen{"$st[0]\0$st[1]"}++;
1178 # favor relative paths for rename-friendliness
1179 push @new, "../../local/$n.git/objects\n";
1181 warn "W: stat($d) failed: $!\n";
1185 # git-multi-pack-index(1) can speed up "git cat-file" startup slightly
1188 my $pd = "$ALL/objects/pack";
1189 if (!mkdir($pd) && $!{EEXIST} && opendir($dh, $pd)) {
1190 # drop stale symlinks
1191 while (defined(my $dn = readdir($dh))) {
1192 if ($dn =~ /\.(?:idx|pack|promisor|bitmap|rev)\z/) {
1194 unlink($f) if -l $f && !-e $f;
1199 for my $ibx (@{ibx_sorted($self, 'active')}) {
1200 # create symlinks for multi-pack-index
1201 $git_midx += symlink_packs($ibx, $pd);
1202 # add new lines to our alternates file
1203 my $line = $ibx->git->{git_dir} . "/objects\n";
1204 chomp(my $d = $line);
1205 if (my @st = stat($d)) {
1206 next if $seen{"$st[0]\0$st[1]"}++;
1208 warn "W: stat($d) failed (from $ibx->{inboxdir}): $!\n";
1209 next if $opt->{-idx_gc};
1216 PublicInbox::V2Writable::write_alternates($info_dir, $mode, $o);
1218 $git_midx and $self->with_umask(sub {
1219 my @cmd = ('multi-pack-index');
1220 push @cmd, '--no-progress' if ($opt->{quiet}//0) > 1;
1221 my $lk = $self->lock_for_scope;
1222 system('git', "--git-dir=$ALL", @cmd, 'write');
1223 # ignore errors, fairly new command, may not exist
1225 $self->parallel_init($self->{indexlevel});
1226 $self->with_umask(\&_idx_init, $self, $opt);
1227 $self->{oidx}->begin_lazy;
1228 $self->{oidx}->eidx_prep;
1229 $self->{midx}->create_xdb if @new;
1232 sub _watch_commit { # PublicInbox::DS::add_timer callback
1234 delete $self->{-commit_timer};
1235 eidxq_process($self, $self->{-watch_sync});
1236 eidxq_release($self);
1237 my $fmt = delete $self->{-watch_sync}->{-regen_fmt};
1238 reindex_checkpoint($self, $self->{-watch_sync});
1239 $self->{-watch_sync}->{-regen_fmt} = $fmt;
1241 # call event_step => done unless commit_timer is armed
1242 PublicInbox::DS::requeue($self);
1245 sub on_inbox_unlock { # called by PublicInbox::InboxIdle
1246 my ($self, $ibx) = @_;
1247 my $opt = $self->{-watch_sync}->{-opt};
1248 my $pr = $opt->{-progress};
1249 my $ekey = $ibx->eidx_key;
1250 local $0 = "sync $ekey";
1251 $pr->("indexing $ekey\n") if $pr;
1252 $self->idx_init($opt);
1253 sync_inbox($self, $self->{-watch_sync}, $ibx);
1254 $self->{-commit_timer} //= add_timer($opt->{'commit-interval'} // 10,
1255 \&_watch_commit, $self);
1258 sub eidx_reload { # -extindex --watch SIGHUP handler
1259 my ($self, $idler) = @_;
1261 my $pr = $self->{-watch_sync}->{-opt}->{-progress};
1262 $pr->('reloading ...') if $pr;
1263 delete $self->{-resync_queue};
1264 delete $self->{-ibx_ary_known};
1265 delete $self->{-ibx_ary_active};
1266 $self->{ibx_known} = [];
1267 $self->{ibx_active} = [];
1268 %{$self->{ibx_map}} = ();
1269 delete $self->{-watch_sync}->{id2pos};
1270 my $cfg = PublicInbox::Config->new;
1271 attach_config($self, $cfg);
1272 $idler->refresh($cfg);
1273 $pr->(" done\n") if $pr;
1275 warn "reload not supported without --all\n";
1279 sub eidx_resync_start ($) { # -extindex --watch SIGUSR1 handler
1281 $self->{-resync_queue} //= [ @{ibx_sorted($self, 'active')} ];
1282 PublicInbox::DS::requeue($self); # trigger our ->event_step
1285 sub event_step { # PublicInbox::DS::requeue callback
1287 if (my $resync_queue = $self->{-resync_queue}) {
1288 if (my $ibx = shift(@$resync_queue)) {
1289 on_inbox_unlock($self, $ibx);
1290 PublicInbox::DS::requeue($self);
1292 delete $self->{-resync_queue};
1293 _watch_commit($self);
1296 done($self) unless $self->{-commit_timer};
1300 sub eidx_watch { # public-inbox-extindex --watch main loop
1301 my ($self, $opt) = @_;
1303 for my $sig (qw(HUP USR1 TSTP QUIT INT TERM)) {
1304 $SIG{$sig} = sub { warn "SIG$sig ignored while scanning\n" };
1306 require PublicInbox::InboxIdle;
1307 require PublicInbox::DS;
1308 require PublicInbox::Syscall;
1309 require PublicInbox::Sigfd;
1310 my $idler = PublicInbox::InboxIdle->new($self->{cfg});
1311 if (!$self->{cfg}) {
1312 $idler->watch_inbox($_) for (@{ibx_sorted($self, 'active')});
1314 for my $ibx (@{ibx_sorted($self, 'active')}) {
1315 $ibx->subscribe_unlock(__PACKAGE__, $self)
1317 my $pr = $opt->{-progress};
1318 $pr->("performing initial scan ...\n") if $pr;
1319 my $sync = eidx_sync($self, $opt); # initial sync
1320 return if $sync->{quit};
1321 my $oldset = PublicInbox::DS::block_signals();
1322 local $self->{current_info} = '';
1323 my $cb = $SIG{__WARN__} || \&CORE::warn;
1324 local $SIG{__WARN__} = sub {
1325 return if PublicInbox::Eml::warn_ignore(@_);
1326 $cb->($self->{current_info}, ': ', @_);
1329 HUP => sub { eidx_reload($self, $idler) },
1330 USR1 => sub { eidx_resync_start($self) },
1331 TSTP => sub { kill('STOP', $$) },
1333 my $quit = PublicInbox::SearchIdx::quit_cb($sync);
1334 $sig->{QUIT} = $sig->{INT} = $sig->{TERM} = $quit;
1335 my $sigfd = PublicInbox::Sigfd->new($sig,
1336 $PublicInbox::Syscall::SFD_NONBLOCK);
1337 %SIG = (%SIG, %$sig) if !$sigfd;
1338 local $self->{-watch_sync} = $sync; # for ->on_inbox_unlock
1340 # wake up every second to accept signals if we don't
1341 # have signalfd or IO::KQueue:
1342 PublicInbox::DS::sig_setmask($oldset);
1343 PublicInbox::DS->SetLoopTimeout(1000);
1345 PublicInbox::DS->SetPostLoopCallback(sub { !$sync->{quit} });
1346 $pr->("initial scan complete, entering event loop\n") if $pr;
1347 PublicInbox::DS->EventLoop; # calls InboxIdle->event_step
1352 *done = \&PublicInbox::V2Writable::done;
1353 *with_umask = \&PublicInbox::InboxWritable::with_umask;
1354 *parallel_init = \&PublicInbox::V2Writable::parallel_init;
1355 *nproc_shards = \&PublicInbox::V2Writable::nproc_shards;
1356 *sync_prepare = \&PublicInbox::V2Writable::sync_prepare;
1357 *index_todo = \&PublicInbox::V2Writable::index_todo;
1358 *count_shards = \&PublicInbox::V2Writable::count_shards;
1359 *atfork_child = \&PublicInbox::V2Writable::atfork_child;
1360 *idx_shard = \&PublicInbox::V2Writable::idx_shard;
1361 *reindex_checkpoint = \&PublicInbox::V2Writable::reindex_checkpoint;
1362 *checkpoint = \&PublicInbox::V2Writable::checkpoint;