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) = @_;
346 my $dbh = $self->{oidx}->dbh;
348 # for debug/info purposes, oids may no longer be accessible
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 $dbh->prepare_cached(<<'')->execute($docid, $ibx_id);
356 DELETE FROM xref3 WHERE docid = ? AND ibx_id = ?
358 my $remain = $self->{oidx}->get_xref3($docid);
359 if (scalar(@$remain)) {
360 $self->{oidx}->eidxq_add($docid); # enqueue for reindex
362 warn "I: unref #$docid $eidx_key $oid\n";
365 warn "I: remove #$docid $eidx_key @oid\n";
366 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
371 my ($self, $opt) = @_;
372 $self->{cfg} or die "E: GC requires ->attach_config\n";
374 $self->idx_init($opt); # acquire lock via V2Writable::_idx_init
376 my $dbh = $self->{oidx}->dbh;
377 $dbh->do('PRAGMA case_sensitive_like = ON'); # only place we use LIKE
378 my $x3_doc = $dbh->prepare('SELECT docid FROM xref3 WHERE ibx_id = ?');
379 my $ibx_ck = $dbh->prepare('SELECT ibx_id,eidx_key FROM inboxes');
380 my $lc_i = $dbh->prepare(<<'');
381 SELECT key FROM eidx_meta WHERE key LIKE ? ESCAPE ?
384 while (my ($ibx_id, $eidx_key) = $ibx_ck->fetchrow_array) {
385 next if $self->{ibx_map}->{$eidx_key};
386 $self->{midx}->remove_eidx_key($eidx_key);
387 warn "I: deleting messages for $eidx_key...\n";
388 $x3_doc->execute($ibx_id);
389 while (defined(my $docid = $x3_doc->fetchrow_array)) {
390 gc_unref_doc($self, $ibx_id, $eidx_key, $docid);
392 $dbh->prepare_cached(<<'')->execute($ibx_id);
393 DELETE FROM inboxes WHERE ibx_id = ?
395 # drop last_commit info
397 $pat =~ s/([_%\\])/\\$1/g;
398 $lc_i->execute("lc-%:$pat//%", '\\');
399 while (my ($key) = $lc_i->fetchrow_array) {
400 next if $key !~ m!\Alc-v[1-9]+:\Q$eidx_key\E//!;
401 warn "I: removing $key\n";
402 $dbh->prepare_cached(<<'')->execute($key);
403 DELETE FROM eidx_meta WHERE key = ?
407 warn "I: $eidx_key removed\n";
410 # it's not real unless it's in `over', we use parallelism here,
411 # shards will be reading directly from over, so commit
412 $self->{oidx}->commit_lazy;
413 $self->{oidx}->begin_lazy;
415 for my $idx (@{$self->{idx_shards}}) {
416 warn "I: cleaning up shard #$idx->{shard}\n";
417 $idx->shard_over_check($self->{oidx});
419 my $nr = $dbh->do(<<'');
420 DELETE FROM xref3 WHERE docid NOT IN (SELECT num FROM over)
422 warn "I: eliminated $nr stale xref3 entries\n" if $nr != 0;
428 my ($self, $sync, $smsg) = @_;
429 my $ibx_id = delete($smsg->{ibx_id}) // die '{ibx_id} unset';
430 my $pos = $sync->{id2pos}->{$ibx_id} // die "$ibx_id no pos";
431 $self->{-ibx_ary_known}->[$pos] //
432 die "BUG: ibx for $smsg->{blob} not mapped"
435 sub _fd_constrained ($) {
437 $self->{-fd_constrained} //= do {
439 if (eval { require BSD::Resource; 1 }) {
440 my $NOFILE = BSD::Resource::RLIMIT_NOFILE();
441 ($soft, undef) = BSD::Resource::getrlimit($NOFILE);
443 chomp($soft = `sh -c 'ulimit -n'`);
445 if (defined($soft)) {
446 # $want is an estimate
447 my $want = scalar(@{$self->{ibx_active}}) + 64;
448 my $ret = $want > $soft;
451 RLIMIT_NOFILE=$soft insufficient (want: $want), will close DB handles early
456 warn "Unable to determine RLIMIT_NOFILE: $@\n";
462 sub _reindex_finalize ($$$) {
463 my ($req, $smsg, $eml) = @_;
464 my $sync = $req->{sync};
465 my $self = $sync->{self};
466 my $by_chash = delete $req->{by_chash} or die 'BUG: no {by_chash}';
467 my $nr = scalar(keys(%$by_chash)) or die 'BUG: no content hashes';
468 my $orig_smsg = $req->{orig_smsg} // die 'BUG: no {orig_smsg}';
469 my $docid = $smsg->{num} = $orig_smsg->{num};
470 $self->{oidx}->add_overview($eml, $smsg); # may rethread
471 check_batch_limit({ %$sync, new_smsg => $smsg });
472 my $chash0 = $smsg->{chash} // die "BUG: $smsg->{blob} no {chash}";
473 my $stable = delete($by_chash->{$chash0}) //
474 die "BUG: $smsg->{blob} chash missing";
475 my $idx = $self->idx_shard($docid);
476 my $top_smsg = pop @$stable;
477 $top_smsg == $smsg or die 'BUG: top_smsg != smsg';
478 my $ibx = _ibx_for($self, $sync, $smsg);
479 $idx->index_eml($eml, $smsg, $ibx->eidx_key);
480 for my $x (reverse @$stable) {
481 $ibx = _ibx_for($self, $sync, $x);
482 my $hdr = delete $x->{hdr} // die 'BUG: no {hdr}';
483 $idx->ipc_do('add_eidx_info', $docid, $ibx->eidx_key, $hdr);
485 return if $nr == 1; # likely, all good
487 warn "W: #$docid split into $nr due to deduplication change\n";
489 for my $ary (values %$by_chash) {
490 for my $x (reverse @$ary) {
491 warn "removing #$docid xref3 $x->{blob}\n";
492 my $n = $self->{oidx}->remove_xref3($docid, $x->{blob});
493 die "BUG: $x->{blob} invalidated #$docid" if $n == 0;
495 my $x = pop(@$ary) // die "BUG: #$docid {by_chash} empty";
496 $x->{num} = delete($x->{xnum}) // die '{xnum} unset';
497 $ibx = _ibx_for($self, $sync, $x);
498 if (my $over = $ibx->over) {
499 my $e = $over->get_art($x->{num});
500 $e->{blob} eq $x->{blob} or die <<EOF;
501 $x->{blob} != $e->{blob} (${\$ibx->eidx_key}:$e->{num});
503 push @todo, $ibx, $e;
504 $over->dbh_close if _fd_constrained($self);
506 die "$ibx->{inboxdir}: over.sqlite3 unusable: $!\n";
510 while (my ($ibx, $e) = splice(@todo, 0, 2)) {
511 reindex_unseen($self, $sync, $ibx, $e);
515 sub _reindex_oid { # git->cat_async callback
516 my ($bref, $oid, $type, $size, $req) = @_;
517 my $sync = $req->{sync};
518 my $self = $sync->{self};
519 my $orig_smsg = $req->{orig_smsg} // die 'BUG: no {orig_smsg}';
520 my $expect_oid = $req->{xr3r}->[$req->{ix}]->[2];
521 my $docid = $orig_smsg->{num};
522 if (is_bad_blob($oid, $type, $size, $expect_oid)) {
523 my $remain = $self->{oidx}->remove_xref3($docid, $expect_oid);
525 warn "W: #$docid gone or corrupted\n";
526 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
527 } elsif (my $next_oid = $req->{xr3r}->[++$req->{ix}]->[2]) {
528 $self->git->cat_async($next_oid, \&_reindex_oid, $req);
530 warn "BUG: #$docid gone (UNEXPECTED)\n";
531 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
535 my $ci = $self->{current_info};
536 local $self->{current_info} = "$ci #$docid $oid";
537 my $re_smsg = bless { blob => $oid }, 'PublicInbox::Smsg';
538 $re_smsg->set_bytes($$bref, $size);
539 my $eml = PublicInbox::Eml->new($bref);
540 $re_smsg->populate($eml, { autime => $orig_smsg->{ds},
541 cotime => $orig_smsg->{ts} });
542 my $chash = content_hash($eml);
543 $re_smsg->{chash} = $chash;
544 $re_smsg->{xnum} = $req->{xr3r}->[$req->{ix}]->[1];
545 $re_smsg->{ibx_id} = $req->{xr3r}->[$req->{ix}]->[0];
546 $re_smsg->{hdr} = $eml->header_obj;
547 push @{$req->{by_chash}->{$chash}}, $re_smsg;
548 if (my $next_oid = $req->{xr3r}->[++$req->{ix}]->[2]) {
549 $self->git->cat_async($next_oid, \&_reindex_oid, $req);
550 } else { # last $re_smsg is the highest priority xref3
551 local $self->{current_info} = "$ci #$docid";
552 _reindex_finalize($req, $re_smsg, $eml);
556 sub _reindex_smsg ($$$) {
557 my ($self, $sync, $smsg) = @_;
558 my $docid = $smsg->{num};
559 my $xr3 = $self->{oidx}->get_xref3($docid, 1);
560 if (scalar(@$xr3) == 0) { # _reindex_check_stale should've covered this
562 BUG? #$docid $smsg->{blob} is not referenced by inboxes during reindex
564 $self->{oidx}->delete_by_num($docid);
565 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
569 # we sort {xr3r} in the reverse order of ibx_sorted so we can
570 # hit the common case in _reindex_finalize without rereading
571 # from git (or holding multiple messages in memory).
572 my $id2pos = $sync->{id2pos}; # index in ibx_sorted
574 $id2pos->{$b->[0]} <=> $id2pos->{$a->[0]}
576 $b->[1] <=> $a->[1] # break ties with {xnum}
578 @$xr3 = map { [ $_->[0], $_->[1], unpack('H*', $_->[2]) ] } @$xr3;
579 my $req = { orig_smsg => $smsg, sync => $sync, xr3r => $xr3, ix => 0 };
580 $self->git->cat_async($xr3->[$req->{ix}]->[2], \&_reindex_oid, $req);
583 sub checkpoint_due ($) {
585 ${$sync->{need_checkpoint}} || (now() > $sync->{next_check});
589 # I've copied FS images and only changed the hostname before,
590 # so prepend hostname. Use `state' since these a BOFH can change
591 # these while this process is running and we always want to be
592 # able to release locks taken by this process.
593 state $retval = hostname . '-' . do {
594 my $m; # machine-id(5) is systemd
595 if (open(my $fh, '<', '/etc/machine-id')) { $m = <$fh> }
596 # (g)hostid(1) is in GNU coreutils, kern.hostid is most BSDs
597 chomp($m ||= `{ sysctl -n kern.hostid ||
598 hostid || ghostid; } 2>/dev/null`
599 || "no-machine-id-or-hostid-on-$^O");
606 my $expect = delete($self->{-eidxq_locked}) or return;
607 my ($owner_pid, undef) = split(/-/, $expect);
608 return if $owner_pid != $$; # shards may fork
609 my $oidx = $self->{oidx};
611 my $cur = $oidx->eidx_meta('eidxq_lock') // '';
612 if ($cur eq $expect) {
613 $oidx->eidx_meta('eidxq_lock', '');
615 } elsif ($cur ne '') {
616 warn "E: eidxq_lock($expect) stolen by $cur\n";
618 warn "E: eidxq_lock($expect) released by another process\n";
625 eidxq_release($self) and $self->{oidx}->commit_lazy;
628 sub _eidxq_take ($) {
630 my $val = "$$-${\time}-$>-".host_ident;
631 $self->{oidx}->eidx_meta('eidxq_lock', $val);
632 $self->{-eidxq_locked} = $val;
635 sub eidxq_lock_acquire ($) {
637 my $oidx = $self->{oidx};
639 my $cur = $oidx->eidx_meta('eidxq_lock') || return _eidxq_take($self);
640 if (my $locked = $self->{-eidxq_locked}) { # be lazy
641 return $locked if $locked eq $cur;
643 my ($pid, $time, $euid, $ident) = split(/-/, $cur, 4);
644 my $t = strftime('%Y-%m-%d %k:%M:%S', gmtime($time));
645 if ($euid == $> && $ident eq host_ident) {
648 I: PID:$pid (re)indexing Xapian since $t, it will continue our work
652 warn "I: eidxq_lock is stale ($cur), clobbering\n";
653 return _eidxq_take($self);
655 warn "E: kill(0, $pid) failed: $!\n"; # fall-through:
657 my $fn = $oidx->dbh->sqlite_db_filename;
659 W: PID:$pid, UID:$euid on $ident is indexing Xapian since $t
660 W: If this is unexpected, delete `eidxq_lock' from the `eidx_meta' table:
661 W: sqlite3 $fn 'DELETE FROM eidx_meta WHERE key = "eidxq_lock"'
666 sub ibx_sorted ($$) {
667 my ($self, $type) = @_;
668 $self->{"-ibx_ary_$type"} //= do {
669 # highest boost first, stable for config-ordering tiebreaker
672 ($b->{boost} // 0) <=> ($a->{boost} // 0)
673 } @{$self->{'ibx_'.$type} // die "BUG: $type unknown"} ];
677 sub prep_id2pos ($) {
681 $id2pos{$_->{-ibx_id}} = $pos++ for (@{ibx_sorted($self, 'known')});
685 sub eidxq_process ($$) { # for reindexing
686 my ($self, $sync) = @_;
688 return unless eidxq_lock_acquire($self);
689 my $dbh = $self->{oidx}->dbh;
690 my $tot = $dbh->selectrow_array('SELECT COUNT(*) FROM eidxq') or return;
692 local $sync->{-regen_fmt} = "%u/$tot\n";
693 my $pr = $sync->{-opt}->{-progress};
695 my $min = $dbh->selectrow_array('SELECT MIN(docid) FROM eidxq');
696 my $max = $dbh->selectrow_array('SELECT MAX(docid) FROM eidxq');
697 $pr->("Xapian indexing $min..$max (total=$tot)\n");
699 $sync->{id2pos} //= prep_id2pos($self);
702 $del = $dbh->prepare('DELETE FROM eidxq WHERE docid = ?');
703 $iter = $dbh->prepare('SELECT docid FROM eidxq ORDER BY docid ASC');
705 while (defined(my $docid = $iter->fetchrow_array)) {
706 last if $sync->{quit};
707 if (my $smsg = $self->{oidx}->get_art($docid)) {
708 _reindex_smsg($self, $sync, $smsg);
710 warn "E: #$docid does not exist in over\n";
712 $del->execute($docid);
715 if (checkpoint_due($sync)) {
716 $dbh = $del = $iter = undef;
717 reindex_checkpoint($self, $sync); # release lock
718 $dbh = $self->{oidx}->dbh;
722 $self->git->async_wait_all;
723 $pr->("reindexed ${$sync->{nr}}/$tot\n") if $pr;
726 sub _reindex_unseen { # git->cat_async callback
727 my ($bref, $oid, $type, $size, $req) = @_;
728 return if is_bad_blob($oid, $type, $size, $req->{oid});
729 my $self = $req->{self} // die 'BUG: {self} unset';
730 local $self->{current_info} = "$self->{current_info} $oid";
731 my $new_smsg = bless { blob => $oid, }, 'PublicInbox::Smsg';
732 $new_smsg->set_bytes($$bref, $size);
733 my $eml = $req->{eml} = PublicInbox::Eml->new($bref);
734 $req->{new_smsg} = $new_smsg;
735 $req->{chash} = content_hash($eml);
736 $req->{mids} = mids($eml); # do_step iterates through this
737 do_step($req); # enter the normal indexing flow
740 # --reindex may catch totally unseen messages, this handles them
741 sub reindex_unseen ($$$$) {
742 my ($self, $sync, $ibx, $xsmsg) = @_;
745 autime => $xsmsg->{ds},
746 cotime => $xsmsg->{ts},
747 oid => $xsmsg->{blob},
749 xnum => $xsmsg->{num},
750 # {mids} and {chash} will be filled in at _reindex_unseen
752 warn "I: reindex_unseen ${\$ibx->eidx_key}:$req->{xnum}:$req->{oid}\n";
753 $self->git->cat_async($xsmsg->{blob}, \&_reindex_unseen, $req);
756 sub _reindex_check_unseen ($$$) {
757 my ($self, $sync, $ibx) = @_;
758 my $ibx_id = $ibx->{-ibx_id};
760 my ($beg, $end) = (1, $slice);
762 # first, check if we missed any messages in target $ibx
764 my $pr = $sync->{-opt}->{-progress};
765 my $ekey = $ibx->eidx_key;
766 local $sync->{-regen_fmt} =
767 "$ekey checking unseen %u/".$ibx->over->max."\n";
770 while (scalar(@{$msgs = $ibx->over->query_xover($beg, $end)})) {
771 ${$sync->{nr}} = $beg;
772 $beg = $msgs->[-1]->{num} + 1;
773 $end = $beg + $slice;
774 if (checkpoint_due($sync)) {
775 reindex_checkpoint($self, $sync); # release lock
778 my $inx3 = $self->{oidx}->dbh->prepare_cached(<<'', undef, 1);
779 SELECT DISTINCT(docid) FROM xref3 WHERE
780 ibx_id = ? AND xnum = ? AND oidbin = ?
782 for my $xsmsg (@$msgs) {
783 my $oidbin = pack('H*', $xsmsg->{blob});
784 $inx3->bind_param(1, $ibx_id);
785 $inx3->bind_param(2, $xsmsg->{num});
786 $inx3->bind_param(3, $oidbin, SQL_BLOB);
788 my $docids = $inx3->fetchall_arrayref;
789 # index messages which were totally missed
790 # the first time around ASAP:
791 if (scalar(@$docids) == 0) {
792 reindex_unseen($self, $sync, $ibx, $xsmsg);
793 } else { # already seen, reindex later
794 for my $r (@$docids) {
795 $self->{oidx}->eidxq_add($r->[0]);
798 last if $sync->{quit};
800 last if $sync->{quit};
804 sub _reindex_check_stale ($$$) {
805 my ($self, $sync, $ibx) = @_;
807 my $pr = $sync->{-opt}->{-progress};
809 my $ekey = $ibx->eidx_key;
810 local $sync->{-regen_fmt} =
811 "$ekey check stale/missing %u/".$ibx->over->max."\n";
814 if (checkpoint_due($sync)) {
815 reindex_checkpoint($self, $sync); # release lock
817 # now, check if there's stale xrefs
818 my $iter = $self->{oidx}->dbh->prepare_cached(<<'', undef, 1);
819 SELECT docid,xnum,oidbin FROM xref3 WHERE ibx_id = ? AND docid > ?
820 ORDER BY docid,xnum ASC LIMIT 10000
822 $iter->execute($ibx->{-ibx_id}, $min);
825 while (my ($docid, $xnum, $oidbin) = $iter->fetchrow_array) {
826 return if $sync->{quit};
827 ${$sync->{nr}} = $xnum;
829 $fetching = $min = $docid;
830 my $smsg = $ibx->over->get_art($xnum);
834 } elsif (pack('H*', $smsg->{blob}) ne $oidbin) {
835 $err = "mismatch (!= $smsg->{blob})";
837 next; # likely, all good
839 # current_info already has eidx_key
840 my $oidhex = unpack('H*', $oidbin);
841 warn "$xnum:$oidhex (#$docid): $err\n";
842 my $del = $self->{oidx}->dbh->prepare_cached(<<'');
843 DELETE FROM xref3 WHERE ibx_id = ? AND xnum = ? AND oidbin = ?
845 $del->bind_param(1, $ibx->{-ibx_id});
846 $del->bind_param(2, $xnum);
847 $del->bind_param(3, $oidbin, SQL_BLOB);
850 # get_xref3 over-fetches, but this is a rare path:
851 my $xr3 = $self->{oidx}->get_xref3($docid);
852 my $idx = $self->idx_shard($docid);
853 if (scalar(@$xr3) == 0) { # all gone
854 $self->{oidx}->delete_by_num($docid);
855 $self->{oidx}->eidxq_del($docid);
856 $idx->ipc_do('xdb_remove', $docid);
857 } else { # enqueue for reindex of remaining messages
858 $idx->ipc_do('remove_eidx_info',
859 $docid, $ibx->eidx_key);
860 $self->{oidx}->eidxq_add($docid); # yes, add
863 } while (defined $fetching);
866 sub _reindex_inbox ($$$) {
867 my ($self, $sync, $ibx) = @_;
868 my $ekey = $ibx->eidx_key;
869 local $self->{current_info} = $ekey;
870 if (defined(my $err = _ibx_index_reject($ibx))) {
871 warn "W: cannot reindex $ekey ($err)\n";
873 _reindex_check_unseen($self, $sync, $ibx);
874 _reindex_check_stale($self, $sync, $ibx) unless $sync->{quit};
876 delete @$ibx{qw(over mm search git)}; # won't need these for a bit
880 my ($self, $sync) = @_;
882 # acquire eidxq_lock early because full reindex takes forever
883 # and incremental -extindex processes can run during our checkpoints
884 if (!eidxq_lock_acquire($self)) {
885 warn "E: aborting --reindex\n";
888 for my $ibx (@{ibx_sorted($self, 'active')}) {
889 _reindex_inbox($self, $sync, $ibx);
890 last if $sync->{quit};
892 $self->git->async_wait_all; # ensure eidxq gets filled completely
893 eidxq_process($self, $sync) unless $sync->{quit};
897 my ($self, $sync, $ibx) = @_;
898 my $err = _sync_inbox($self, $sync, $ibx);
899 delete @$ibx{qw(mm over)};
900 warn $err, "\n" if defined($err);
903 sub dd_smsg { # git->cat_async callback
904 my ($bref, $oid, $type, $size, $dd) = @_;
905 my $smsg = $dd->{smsg} // die 'BUG: dd->{smsg} missing';
906 my $self = $dd->{self} // die 'BUG: {self} missing';
907 my $per_mid = $dd->{per_mid} // die 'BUG: {per_mid} missing';
908 if ($type eq 'missing') {
909 _blob_missing($dd, $smsg);
910 } elsif (!is_bad_blob($oid, $type, $size, $smsg->{blob})) {
911 local $self->{current_info} = "$self->{current_info} $oid";
912 my $chash = content_hash(PublicInbox::Eml->new($bref));
913 push(@{$per_mid->{dd_chash}->{$chash}}, $smsg);
915 return if $per_mid->{last_smsg} != $smsg;
916 while (my ($chash, $ary) = each %{$per_mid->{dd_chash}}) {
917 my $keep = shift @$ary;
918 next if !scalar(@$ary);
919 $per_mid->{sync}->{dedupe_cull} += scalar(@$ary);
921 "# <$keep->{mid}> keeping #$keep->{num}, dropping ",
922 join(', ', map { "#$_->{num}" } @$ary),"\n";
923 next if $per_mid->{sync}->{-opt}->{'dry-run'};
924 my $oidx = $self->{oidx};
925 for my $smsg (@$ary) {
926 my $gone = $smsg->{num};
927 $oidx->merge_xref3($keep->{num}, $gone, $smsg->{blob});
928 $self->idx_shard($gone)->ipc_do('xdb_remove', $gone);
929 $oidx->delete_by_num($gone);
934 sub eidx_dedupe ($$$) {
935 my ($self, $sync, $msgids) = @_;
936 $sync->{dedupe_cull} = 0;
939 return unless eidxq_lock_acquire($self);
940 my ($iter, $cur_mid);
943 my ($max_id) = $self->{oidx}->dbh->selectrow_array(<<EOS);
944 SELECT MAX(id) FROM msgid
946 local $sync->{-regen_fmt} = "dedupe %u/$max_id\n";
948 # note: we could write this query more intelligently,
949 # but that causes lock contention with read-only processes
951 $cur_mid = $msgids->[$idx];
952 if ($cur_mid eq '') { # all Message-IDs
953 $iter = $self->{oidx}->dbh->prepare(<<EOS);
954 SELECT mid,id FROM msgid WHERE id > ? ORDER BY id ASC
956 $iter->execute($min_id);
958 $iter = $self->{oidx}->dbh->prepare(<<EOS);
959 SELECT mid,id FROM msgid WHERE mid = ? AND id > ? ORDER BY id ASC
961 $iter->execute($cur_mid, $min_id);
963 while (my ($mid, $id) = $iter->fetchrow_array) {
964 last if $sync->{quit};
965 $self->{current_info} = "dedupe $mid";
966 ${$sync->{nr}} = $min_id = $id;
968 while (my $x = $self->{oidx}->next_by_mid($mid, \$id, \$prv)) {
971 next if scalar(@smsg) < 2;
973 dd_chash => {}, # chash => [ary of smsgs]
974 last_smsg => $smsg[-1],
978 $candidates += scalar(@smsg) - 1;
979 for my $smsg (@smsg) {
985 $self->git->cat_async($smsg->{blob}, \&dd_smsg, $dd);
987 # need to wait on every single one @smsg contents can get
988 # invalidated inside dd_smsg for messages with multiple
990 $self->git->async_wait_all;
992 if (checkpoint_due($sync)) {
994 reindex_checkpoint($self, $sync);
998 goto dedupe_restart if defined($msgids->[++$idx]);
1000 my $n = delete $sync->{dedupe_cull};
1001 if (my $pr = $sync->{-opt}->{-progress}) {
1002 $pr->("culled $n/$candidates candidates ($nr_mid msgids)\n");
1007 sub eidx_sync { # main entry point
1008 my ($self, $opt) = @_;
1010 my $warn_cb = $SIG{__WARN__} || \&CORE::warn;
1011 local $self->{current_info} = '';
1012 local $SIG{__WARN__} = sub {
1013 return if PublicInbox::Eml::warn_ignore(@_);
1014 $warn_cb->($self->{current_info}, ': ', @_);
1016 $self->idx_init($opt); # acquire lock via V2Writable::_idx_init
1017 $self->{oidx}->rethread_prepare($opt);
1019 need_checkpoint => \(my $need_checkpoint = 0),
1021 next_check => now() + 10,
1023 # DO NOT SET {reindex} here, it's incompatible with reused
1024 # V2Writable code, reindex is totally different here
1025 # compared to v1/v2 inboxes because we have multiple histories
1027 -regen_fmt => "%u/?\n",
1029 local $SIG{USR1} = sub { $need_checkpoint = 1 };
1030 my $quit = PublicInbox::SearchIdx::quit_cb($sync);
1031 local $SIG{QUIT} = $quit;
1032 local $SIG{INT} = $quit;
1033 local $SIG{TERM} = $quit;
1034 for my $ibx (@{ibx_sorted($self, 'known')}) {
1035 $ibx->{-ibx_id} //= $self->{oidx}->ibx_id($ibx->eidx_key);
1038 if (scalar(grep { defined($_->{boost}) } @{$self->{ibx_known}})) {
1039 $sync->{id2pos} //= prep_id2pos($self);
1040 $sync->{boost_in_use} = 1;
1043 if (my $msgids = delete($opt->{dedupe})) {
1044 local $sync->{checkpoint_unlocks} = 1;
1045 eidx_dedupe($self, $sync, $msgids);
1047 if (delete($opt->{reindex})) {
1048 local $sync->{checkpoint_unlocks} = 1;
1049 eidx_reindex($self, $sync);
1052 # don't use $_ here, it'll get clobbered by reindex_checkpoint
1053 if ($opt->{scan} // 1) {
1054 for my $ibx (@{ibx_sorted($self, 'active')}) {
1055 last if $sync->{quit};
1056 sync_inbox($self, $sync, $ibx);
1059 $self->{oidx}->rethread_done($opt) unless $sync->{quit};
1060 eidxq_process($self, $sync) unless $sync->{quit};
1062 eidxq_release($self);
1064 $sync; # for eidx_watch
1067 sub update_last_commit { # overrides V2Writable
1068 my ($self, $sync, $stk) = @_;
1069 my $unit = $sync->{unit} // return;
1070 my $latest_cmt = $stk ? $stk->{latest_cmt} : ${$sync->{latest_cmt}};
1071 defined($latest_cmt) or return;
1072 my $ibx = $sync->{ibx} or die 'BUG: {ibx} missing';
1073 my $ekey = $ibx->eidx_key;
1074 my $uv = $ibx->uidvalidity;
1075 my $epoch = $unit->{epoch};
1077 my $v = $ibx->version;
1079 die 'No {epoch} for v2 unit' unless defined $epoch;
1080 $meta_key = "lc-v2:$ekey//$uv;$epoch";
1082 die 'Unexpected {epoch} for v1 unit' if defined $epoch;
1083 $meta_key = "lc-v1:$ekey//$uv";
1085 die "Unsupported inbox version: $v";
1087 my $last = $self->{oidx}->eidx_meta($meta_key);
1088 if (defined $last && is_ancestor($self->git, $last, $latest_cmt)) {
1089 my @cmd = (qw(rev-list --count), "$last..$latest_cmt");
1090 chomp(my $n = $unit->{git}->qx(@cmd));
1091 return if $n ne '' && $n == 0;
1093 $self->{oidx}->eidx_meta($meta_key, $latest_cmt);
1096 sub _idx_init { # with_umask callback
1097 my ($self, $opt) = @_;
1098 PublicInbox::V2Writable::_idx_init($self, $opt); # acquires ei.lock
1099 $self->{midx} = PublicInbox::MiscIdx->new($self);
1102 sub symlink_packs ($$) {
1103 my ($ibx, $pd) = @_;
1105 my $glob = "$ibx->{inboxdir}/git/*.git/objects/pack/*.idx";
1106 for my $idx (bsd_glob($glob, GLOB_NOSORT)) {
1107 my $src = substr($idx, 0, -length('.idx'));
1108 my $dst = $pd . substr($src, rindex($src, '/'));
1109 if (-f "$src.pack" and
1110 symlink("$src.pack", "$dst.pack") and
1111 symlink($idx, "$dst.idx") and
1114 # .promisor, .bitmap, .rev and .keep are optional
1115 # XXX should we symlink .keep here?
1116 for my $s (qw(promisor bitmap rev)) {
1117 symlink("$src.$s", "$dst.$s") if -f "$src.$s";
1119 } elsif (!$!{EEXIST}) {
1120 warn "W: ln -s $src.{pack,idx} => $dst.*: $!\n";
1121 unlink "$dst.pack", "$dst.idx";
1127 sub idx_init { # similar to V2Writable
1128 my ($self, $opt) = @_;
1129 return if $self->{idx_shards};
1131 $self->git->cleanup;
1133 my $ALL = $self->git->{git_dir}; # ALL.git
1135 if ($opt->{-private}) { # LeiStore
1138 umask 077; # don't bother restoring
1139 PublicInbox::Import::init_bare($ALL);
1140 $self->git->qx(qw(config core.sharedRepository 0600));
1143 PublicInbox::Import::init_bare($ALL) unless $old;
1145 my $info_dir = "$ALL/objects/info";
1146 my $alt = "$info_dir/alternates";
1147 my (@old, @new, %seen); # seen: st_dev + st_ino
1149 open(my $fh, '<', $alt) or die "open $alt: $!";
1150 $mode = (stat($fh))[2] & 07777;
1151 while (my $line = <$fh>) {
1152 chomp(my $d = $line);
1154 # expand relative path (/local/ stuff)
1155 substr($d, 0, 3) eq '../' and
1156 $d = "$ALL/objects/$d";
1157 if (my @st = stat($d)) {
1158 next if $seen{"$st[0]\0$st[1]"}++;
1160 warn "W: stat($d) failed (from $alt): $!\n";
1161 next if $opt->{-idx_gc};
1167 # for LeiStore, and possibly some mirror-only state
1168 if (opendir(my $dh, my $local = "$self->{topdir}/local")) {
1169 # highest numbered epoch first
1170 for my $n (sort { $b <=> $a } map { substr($_, 0, -4) + 0 }
1171 grep(/\A[0-9]+\.git\z/, readdir($dh))) {
1172 my $d = "$local/$n.git/objects"; # absolute path
1173 if (my @st = stat($d)) {
1174 next if $seen{"$st[0]\0$st[1]"}++;
1175 # favor relative paths for rename-friendliness
1176 push @new, "../../local/$n.git/objects\n";
1178 warn "W: stat($d) failed: $!\n";
1182 # git-multi-pack-index(1) can speed up "git cat-file" startup slightly
1185 my $pd = "$ALL/objects/pack";
1186 if (!mkdir($pd) && $!{EEXIST} && opendir($dh, $pd)) {
1187 # drop stale symlinks
1188 while (defined(my $dn = readdir($dh))) {
1189 if ($dn =~ /\.(?:idx|pack|promisor|bitmap|rev)\z/) {
1191 unlink($f) if -l $f && !-e $f;
1196 for my $ibx (@{ibx_sorted($self, 'active')}) {
1197 # create symlinks for multi-pack-index
1198 $git_midx += symlink_packs($ibx, $pd);
1199 # add new lines to our alternates file
1200 my $line = $ibx->git->{git_dir} . "/objects\n";
1201 chomp(my $d = $line);
1202 if (my @st = stat($d)) {
1203 next if $seen{"$st[0]\0$st[1]"}++;
1205 warn "W: stat($d) failed (from $ibx->{inboxdir}): $!\n";
1206 next if $opt->{-idx_gc};
1213 PublicInbox::V2Writable::write_alternates($info_dir, $mode, $o);
1215 $git_midx and $self->with_umask(sub {
1216 my @cmd = ('multi-pack-index');
1217 push @cmd, '--no-progress' if ($opt->{quiet}//0) > 1;
1218 my $lk = $self->lock_for_scope;
1219 system('git', "--git-dir=$ALL", @cmd, 'write');
1220 # ignore errors, fairly new command, may not exist
1222 $self->parallel_init($self->{indexlevel});
1223 $self->with_umask(\&_idx_init, $self, $opt);
1224 $self->{oidx}->begin_lazy;
1225 $self->{oidx}->eidx_prep;
1226 $self->{midx}->create_xdb if @new;
1229 sub _watch_commit { # PublicInbox::DS::add_timer callback
1231 delete $self->{-commit_timer};
1232 eidxq_process($self, $self->{-watch_sync});
1233 eidxq_release($self);
1234 my $fmt = delete $self->{-watch_sync}->{-regen_fmt};
1235 reindex_checkpoint($self, $self->{-watch_sync});
1236 $self->{-watch_sync}->{-regen_fmt} = $fmt;
1238 # call event_step => done unless commit_timer is armed
1239 PublicInbox::DS::requeue($self);
1242 sub on_inbox_unlock { # called by PublicInbox::InboxIdle
1243 my ($self, $ibx) = @_;
1244 my $opt = $self->{-watch_sync}->{-opt};
1245 my $pr = $opt->{-progress};
1246 my $ekey = $ibx->eidx_key;
1247 local $0 = "sync $ekey";
1248 $pr->("indexing $ekey\n") if $pr;
1249 $self->idx_init($opt);
1250 sync_inbox($self, $self->{-watch_sync}, $ibx);
1251 $self->{-commit_timer} //= add_timer($opt->{'commit-interval'} // 10,
1252 \&_watch_commit, $self);
1255 sub eidx_reload { # -extindex --watch SIGHUP handler
1256 my ($self, $idler) = @_;
1258 my $pr = $self->{-watch_sync}->{-opt}->{-progress};
1259 $pr->('reloading ...') if $pr;
1260 delete $self->{-resync_queue};
1261 delete $self->{-ibx_ary_known};
1262 delete $self->{-ibx_ary_active};
1263 $self->{ibx_known} = [];
1264 $self->{ibx_active} = [];
1265 %{$self->{ibx_map}} = ();
1266 delete $self->{-watch_sync}->{id2pos};
1267 my $cfg = PublicInbox::Config->new;
1268 attach_config($self, $cfg);
1269 $idler->refresh($cfg);
1270 $pr->(" done\n") if $pr;
1272 warn "reload not supported without --all\n";
1276 sub eidx_resync_start ($) { # -extindex --watch SIGUSR1 handler
1278 $self->{-resync_queue} //= [ @{ibx_sorted($self, 'active')} ];
1279 PublicInbox::DS::requeue($self); # trigger our ->event_step
1282 sub event_step { # PublicInbox::DS::requeue callback
1284 if (my $resync_queue = $self->{-resync_queue}) {
1285 if (my $ibx = shift(@$resync_queue)) {
1286 on_inbox_unlock($self, $ibx);
1287 PublicInbox::DS::requeue($self);
1289 delete $self->{-resync_queue};
1290 _watch_commit($self);
1293 done($self) unless $self->{-commit_timer};
1297 sub eidx_watch { # public-inbox-extindex --watch main loop
1298 my ($self, $opt) = @_;
1300 for my $sig (qw(HUP USR1 TSTP QUIT INT TERM)) {
1301 $SIG{$sig} = sub { warn "SIG$sig ignored while scanning\n" };
1303 require PublicInbox::InboxIdle;
1304 require PublicInbox::DS;
1305 require PublicInbox::Syscall;
1306 require PublicInbox::Sigfd;
1307 my $idler = PublicInbox::InboxIdle->new($self->{cfg});
1308 if (!$self->{cfg}) {
1309 $idler->watch_inbox($_) for (@{ibx_sorted($self, 'active')});
1311 for my $ibx (@{ibx_sorted($self, 'active')}) {
1312 $ibx->subscribe_unlock(__PACKAGE__, $self)
1314 my $pr = $opt->{-progress};
1315 $pr->("performing initial scan ...\n") if $pr;
1316 my $sync = eidx_sync($self, $opt); # initial sync
1317 return if $sync->{quit};
1318 my $oldset = PublicInbox::DS::block_signals();
1319 local $self->{current_info} = '';
1320 my $cb = $SIG{__WARN__} || \&CORE::warn;
1321 local $SIG{__WARN__} = sub {
1322 return if PublicInbox::Eml::warn_ignore(@_);
1323 $cb->($self->{current_info}, ': ', @_);
1326 HUP => sub { eidx_reload($self, $idler) },
1327 USR1 => sub { eidx_resync_start($self) },
1328 TSTP => sub { kill('STOP', $$) },
1330 my $quit = PublicInbox::SearchIdx::quit_cb($sync);
1331 $sig->{QUIT} = $sig->{INT} = $sig->{TERM} = $quit;
1332 my $sigfd = PublicInbox::Sigfd->new($sig,
1333 $PublicInbox::Syscall::SFD_NONBLOCK);
1334 %SIG = (%SIG, %$sig) if !$sigfd;
1335 local $self->{-watch_sync} = $sync; # for ->on_inbox_unlock
1337 # wake up every second to accept signals if we don't
1338 # have signalfd or IO::KQueue:
1339 PublicInbox::DS::sig_setmask($oldset);
1340 PublicInbox::DS->SetLoopTimeout(1000);
1342 PublicInbox::DS->SetPostLoopCallback(sub { !$sync->{quit} });
1343 $pr->("initial scan complete, entering event loop\n") if $pr;
1344 PublicInbox::DS->EventLoop; # calls InboxIdle->event_step
1349 *done = \&PublicInbox::V2Writable::done;
1350 *with_umask = \&PublicInbox::InboxWritable::with_umask;
1351 *parallel_init = \&PublicInbox::V2Writable::parallel_init;
1352 *nproc_shards = \&PublicInbox::V2Writable::nproc_shards;
1353 *sync_prepare = \&PublicInbox::V2Writable::sync_prepare;
1354 *index_todo = \&PublicInbox::V2Writable::index_todo;
1355 *count_shards = \&PublicInbox::V2Writable::count_shards;
1356 *atfork_child = \&PublicInbox::V2Writable::atfork_child;
1357 *idx_shard = \&PublicInbox::V2Writable::idx_shard;
1358 *reindex_checkpoint = \&PublicInbox::V2Writable::reindex_checkpoint;
1359 *checkpoint = \&PublicInbox::V2Writable::checkpoint;