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::MultiGit;
25 use PublicInbox::Search;
26 use PublicInbox::SearchIdx qw(prepare_stack is_ancestor is_bad_blob);
27 use PublicInbox::OverIdx;
28 use PublicInbox::MiscIdx;
29 use PublicInbox::MID qw(mids);
30 use PublicInbox::V2Writable;
31 use PublicInbox::InboxWritable;
32 use PublicInbox::ContentHash qw(content_hash);
34 use PublicInbox::DS qw(now add_timer);
35 use DBI qw(:sql_types); # SQL_BLOB
38 my (undef, $dir, $opt) = @_;
39 my $l = $opt->{indexlevel} // 'full';
40 $l !~ $PublicInbox::SearchIdx::INDEXLEVELS and
41 die "invalid indexlevel=$l\n";
42 $l eq 'basic' and die "E: indexlevel=basic not yet supported\n";
44 xpfx => "$dir/ei".PublicInbox::Search::SCHEMA_VERSION,
46 creat => $opt->{creat},
47 ibx_map => {}, # (newsgroup//inboxdir) => $ibx
48 ibx_active => [], # by config section order
49 ibx_known => [], # by config section order
55 lock_path => "$dir/ei.lock",
57 $self->{shards} = $self->count_shards ||
58 nproc_shards({ nproc => $opt->{jobs} });
59 my $oidx = PublicInbox::OverIdx->new("$self->{xpfx}/over.sqlite3");
60 $self->{-no_fsync} = $oidx->{-no_fsync} = 1 if !$opt->{fsync};
61 $self->{oidx} = $oidx;
66 my ($self, $ibx, $types) = @_;
67 $self->{ibx_map}->{$ibx->eidx_key} //= do {
68 delete $self->{-ibx_ary_known}; # invalidate cache
69 delete $self->{-ibx_ary_active}; # invalidate cache
70 $types //= [ qw(active known) ];
72 push @{$self->{"ibx_$t"}}, $ibx;
78 sub _ibx_attach { # each_inbox callback
79 my ($ibx, $self, $types) = @_;
80 attach_inbox($self, $ibx, $types);
84 my ($self, $cfg, $ibxs) = @_;
88 for my $ibx (@$ibxs) {
89 $self->{ibx_map}->{$ibx->eidx_key} //= do {
90 push @{$self->{ibx_active}}, $ibx;
91 push @{$self->{ibx_known}}, $ibx;
96 delete $self->{-ibx_ary_known};
97 delete $self->{-ibx_ary_active};
100 $types //= [ qw(known active) ];
101 $cfg->each_inbox(\&_ibx_attach, $self, $types);
104 sub check_batch_limit ($) {
106 my $self = $req->{self};
107 my $new_smsg = $req->{new_smsg};
108 my $n = $self->{transact_bytes} += $new_smsg->{bytes};
110 # set flag for PublicInbox::V2Writable::index_todo:
111 ${$req->{need_checkpoint}} = 1 if $n >= $self->{batch_bytes};
114 sub apply_boost ($$) {
115 my ($req, $smsg) = @_;
116 my $id2pos = $req->{id2pos}; # index in ibx_sorted
117 my $xr3 = $req->{self}->{oidx}->get_xref3($smsg->{num}, 1);
119 $id2pos->{$a->[0]} <=> $id2pos->{$b->[0]}
121 $a->[1] <=> $b->[1] # break ties with {xnum}
123 my $new_smsg = $req->{new_smsg};
124 return if $xr3->[0]->[2] ne pack('H*', $new_smsg->{blob}); # loser
126 # replace the old smsg with the more boosted one
127 $new_smsg->{num} = $smsg->{num};
128 $new_smsg->populate($req->{eml}, $req);
129 $req->{self}->{oidx}->add_overview($req->{eml}, $new_smsg);
132 sub _unref_doc ($$$$$;$) {
133 my ($sync, $docid, $ibx, $xnum, $oidbin, $eml) = @_;
134 my $s = 'DELETE FROM xref3 WHERE ibx_id = ? AND oidbin = ?';
135 $s .= ' AND xnum = ?' if defined($xnum);
136 my $del = $sync->{self}->{oidx}->dbh->prepare_cached($s);
137 $del->bind_param(1, $ibx->{-ibx_id});
138 $del->bind_param(2, $oidbin, SQL_BLOB);
139 $del->bind_param(3, $xnum) if defined($xnum);
141 my $xr3 = $sync->{self}->{oidx}->get_xref3($docid, 1);
142 my $idx = $sync->{self}->idx_shard($docid);
143 if (scalar(@$xr3) == 0) { # all gone
144 $sync->{self}->{oidx}->delete_by_num($docid);
145 $sync->{self}->{oidx}->eidxq_del($docid);
146 $idx->ipc_do('xdb_remove', $docid);
147 } else { # enqueue for reindex of remaining messages
148 my $ekey = $ibx->{-gc_eidx_key} // $ibx->eidx_key;
149 $idx->ipc_do('remove_eidx_info', $docid, $ekey, $eml);
150 $sync->{self}->{oidx}->eidxq_add($docid); # yes, add
156 my ($req, $smsg) = @_;
157 my $self = $req->{self};
158 my $docid = $smsg->{num};
159 my $oid = $req->{oid};
160 my $xibx = $req->{ibx};
161 my $eml = $req->{eml};
162 if (my $new_smsg = $req->{new_smsg}) { # 'm' on cross-posted message
163 my $eidx_key = $xibx->eidx_key;
164 my $xnum = $req->{xnum};
165 $self->{oidx}->add_xref3($docid, $xnum, $oid, $eidx_key);
166 my $idx = $self->idx_shard($docid);
167 $idx->ipc_do('add_eidx_info', $docid, $eidx_key, $eml);
168 apply_boost($req, $smsg) if $req->{boost_in_use};
169 } else { # 'd' no {xnum}
170 $oid = pack('H*', $oid);
171 _unref_doc($req, $docid, $xibx, undef, $oid, $eml);
175 # called by V2Writable::sync_prepare
176 sub artnum_max { $_[0]->{oidx}->eidx_max }
178 sub index_unseen ($) {
180 my $new_smsg = $req->{new_smsg} or die 'BUG: {new_smsg} unset';
181 my $eml = delete $req->{eml};
182 $new_smsg->populate($eml, $req);
183 my $self = $req->{self};
184 my $docid = $self->{oidx}->adj_counter('eidx_docid', '+');
185 $new_smsg->{num} = $docid;
186 my $idx = $self->idx_shard($docid);
187 $self->{oidx}->add_overview($eml, $new_smsg);
188 my $oid = $new_smsg->{blob};
189 my $ibx = delete $req->{ibx} or die 'BUG: {ibx} unset';
190 $self->{oidx}->add_xref3($docid, $req->{xnum}, $oid, $ibx->eidx_key);
191 $idx->index_eml($eml, $new_smsg, $ibx->eidx_key);
192 check_batch_limit($req);
195 sub do_finalize ($) {
197 if (my $indexed = $req->{indexed}) { # duplicated messages
198 do_xpost($req, $_) for @$indexed;
199 } elsif (exists $req->{new_smsg}) { # totally unseen messsage
202 # `d' message was already unindexed in the v1/v2 inboxes,
203 # so it's too noisy to warn, here.
205 # cur_cmt may be undef for unindex_oid, set by V2Writable::index_todo
206 if (defined(my $cur_cmt = $req->{cur_cmt})) {
207 ${$req->{latest_cmt}} = $cur_cmt;
211 sub do_step ($) { # main iterator for adding messages to the index
213 my $self = $req->{self} // die 'BUG: {self} missing';
215 if (my $next_arg = $req->{next_arg}) {
216 if (my $smsg = $self->{oidx}->next_by_mid(@$next_arg)) {
217 $req->{cur_smsg} = $smsg;
218 $self->git->cat_async($smsg->{blob},
219 \&ck_existing, $req);
220 return; # ck_existing calls do_step
222 delete $req->{next_arg};
224 die "BUG: {cur_smsg} still set" if $req->{cur_smsg};
225 my $mid = shift(@{$req->{mids}}) // last;
227 $req->{next_arg} = [ $mid, \$id, \$prev ];
233 sub _blob_missing ($$) { # called when $smsg->{blob} is bad
234 my ($req, $smsg) = @_;
235 my $self = $req->{self};
236 my $xref3 = $self->{oidx}->get_xref3($smsg->{num});
237 my @keep = grep(!/:$smsg->{blob}\z/, @$xref3);
239 warn "E: $smsg->{blob} gone, removing #$smsg->{num}\n";
240 $keep[0] =~ /:([a-f0-9]{40,}+)\z/ or
241 die "BUG: xref $keep[0] has no OID";
243 $self->{oidx}->remove_xref3($smsg->{num}, $smsg->{blob});
244 $self->{oidx}->update_blob($smsg, $oidhex) or warn <<EOM;
245 E: #$smsg->{num} gone ($smsg->{blob} => $oidhex)
248 warn "E: $smsg->{blob} gone, removing #$smsg->{num}\n";
249 $self->{oidx}->delete_by_num($smsg->{num});
253 sub ck_existing { # git->cat_async callback
254 my ($bref, $oid, $type, $size, $req) = @_;
255 my $smsg = delete $req->{cur_smsg} or die 'BUG: {cur_smsg} missing';
256 if ($type eq 'missing') {
257 _blob_missing($req, $smsg);
258 } elsif (!is_bad_blob($oid, $type, $size, $smsg->{blob})) {
259 my $self = $req->{self} // die 'BUG: {self} missing';
260 local $self->{current_info} = "$self->{current_info} $oid";
261 my $cur = PublicInbox::Eml->new($bref);
262 if (content_hash($cur) eq $req->{chash}) {
263 push @{$req->{indexed}}, $smsg; # for do_xpost
264 } # else { index_unseen later }
269 # is the messages visible in the inbox currently being indexed?
270 # return the number if so
271 sub cur_ibx_xnum ($$) {
272 my ($req, $bref) = @_;
273 my $ibx = $req->{ibx} or die 'BUG: current {ibx} missing';
275 $req->{eml} = PublicInbox::Eml->new($bref);
276 $req->{chash} = content_hash($req->{eml});
277 $req->{mids} = mids($req->{eml});
278 for my $mid (@{$req->{mids}}) {
280 while (my $x = $ibx->over->next_by_mid($mid, \$id, \$prev)) {
281 return $x->{num} if $x->{blob} eq $req->{oid};
287 sub index_oid { # git->cat_async callback for 'm'
288 my ($bref, $oid, $type, $size, $req) = @_;
289 my $self = $req->{self};
290 local $self->{current_info} = "$self->{current_info} $oid";
291 return if is_bad_blob($oid, $type, $size, $req->{oid});
292 my $new_smsg = $req->{new_smsg} = bless {
294 }, 'PublicInbox::Smsg';
295 $new_smsg->set_bytes($$bref, $size);
296 defined($req->{xnum} = cur_ibx_xnum($req, $bref)) or return;
301 sub unindex_oid { # git->cat_async callback for 'd'
302 my ($bref, $oid, $type, $size, $req) = @_;
303 my $self = $req->{self};
304 local $self->{current_info} = "$self->{current_info} $oid";
305 return if is_bad_blob($oid, $type, $size, $req->{oid});
306 return if defined(cur_ibx_xnum($req, $bref)); # was re-added
310 # overrides V2Writable::last_commits, called by sync_ranges via sync_prepare
312 my ($self, $sync) = @_;
314 my $ekey = $sync->{ibx}->eidx_key;
315 my $uv = $sync->{ibx}->uidvalidity;
316 for my $i (0..$sync->{epoch_max}) {
317 $heads->[$i] = $self->{oidx}->eidx_meta("lc-v2:$ekey//$uv;$i");
322 sub _ibx_index_reject ($) {
324 $ibx->mm // return 'unindexed, no msgmap.sqlite3';
325 $ibx->uidvalidity // return 'no UIDVALIDITY';
326 $ibx->over // return 'unindexed, no over.sqlite3';
330 sub _sync_inbox ($$$) {
331 my ($self, $sync, $ibx) = @_;
332 my $ekey = $ibx->eidx_key;
333 if (defined(my $err = _ibx_index_reject($ibx))) {
334 return "W: skipping $ekey ($err)";
337 $sync->{nr} = \(my $nr = 0);
338 my $v = $ibx->version;
340 $sync->{epoch_max} = $ibx->max_git_epoch // return;
341 sync_prepare($self, $sync); # or return # TODO: once MiscIdx is stable
343 my $uv = $ibx->uidvalidity;
344 my $lc = $self->{oidx}->eidx_meta("lc-v1:$ekey//$uv");
345 my $head = $ibx->mm->last_commit //
346 return "E: $ibx->{inboxdir} is not indexed";
347 my $stk = prepare_stack($sync, $lc ? "$lc..$head" : $head);
348 my $unit = { stack => $stk, git => $ibx->git };
349 push @{$sync->{todo}}, $unit;
351 return "E: $ekey unsupported inbox version (v$v)";
353 for my $unit (@{delete($sync->{todo}) // []}) {
354 last if $sync->{quit};
355 index_todo($self, $sync, $unit);
357 $self->{midx}->index_ibx($ibx) unless $sync->{quit};
358 $ibx->git->cleanup; # done with this inbox, now
362 sub eidx_gc_scan_inboxes ($$) {
363 my ($self, $sync) = @_;
364 my ($x3_doc, $ibx_ck);
366 $x3_doc = $self->{oidx}->dbh->prepare(<<EOM);
367 SELECT docid,xnum,oidbin FROM xref3 WHERE ibx_id = ?
369 $ibx_ck = $self->{oidx}->dbh->prepare(<<EOM);
370 SELECT ibx_id,eidx_key FROM inboxes
373 while (my ($ibx_id, $eidx_key) = $ibx_ck->fetchrow_array) {
374 next if $self->{ibx_map}->{$eidx_key};
375 $self->{midx}->remove_eidx_key($eidx_key);
376 warn "I: deleting messages for $eidx_key...\n";
377 $x3_doc->execute($ibx_id);
378 my $ibx = { -ibx_id => $ibx_id, -gc_eidx_key => $eidx_key };
379 while (my ($docid, $xnum, $oid) = $x3_doc->fetchrow_array) {
380 my $r = _unref_doc($sync, $docid, $ibx, $xnum, $oid);
381 $oid = unpack('H*', $oid);
382 $r = $r ? 'unref' : 'remove';
383 warn "I: $r #$docid $eidx_key $oid\n";
384 if (checkpoint_due($sync)) {
385 $x3_doc = $ibx_ck = undef;
386 reindex_checkpoint($self, $sync);
390 $self->{oidx}->dbh->do(<<'', undef, $ibx_id);
391 DELETE FROM inboxes WHERE ibx_id = ?
393 # drop last_commit info
395 $pat =~ s/([_%\\])/\\$1/g;
396 $self->{oidx}->dbh->do('PRAGMA case_sensitive_like = ON');
397 my $lc_i = $self->{oidx}->dbh->prepare(<<'');
398 SELECT key FROM eidx_meta WHERE key LIKE ? ESCAPE ?
400 $lc_i->execute("lc-%:$pat//%", '\\');
401 while (my ($key) = $lc_i->fetchrow_array) {
402 next if $key !~ m!\Alc-v[1-9]+:\Q$eidx_key\E//!;
403 warn "I: removing $key\n";
404 $self->{oidx}->dbh->do(<<'', undef, $key);
405 DELETE FROM eidx_meta WHERE key = ?
408 warn "I: $eidx_key removed\n";
412 sub eidx_gc_scan_shards ($$) { # TODO: use for lei/store
413 my ($self, $sync) = @_;
414 my $nr = $self->{oidx}->dbh->do(<<'');
415 DELETE FROM xref3 WHERE docid NOT IN (SELECT num FROM over)
417 warn "I: eliminated $nr stale xref3 entries\n" if $nr != 0;
418 reindex_checkpoint($self, $sync) if checkpoint_due($sync);
420 # fixup from old bugs:
421 $nr = $self->{oidx}->dbh->do(<<'');
422 DELETE FROM over WHERE num > 0 AND num NOT IN (SELECT docid FROM xref3)
424 warn "I: eliminated $nr stale over entries\n" if $nr != 0;
425 reindex_checkpoint($self, $sync) if checkpoint_due($sync);
427 my ($cur) = $self->{oidx}->dbh->selectrow_array(<<EOM);
428 SELECT MIN(num) FROM over WHERE num > 0
430 $cur // return; # empty
431 my ($r, $n, %active_shards);
434 $r = $self->{oidx}->dbh->selectcol_arrayref(<<"", undef, $cur);
435 SELECT num FROM over WHERE num >= ? ORDER BY num ASC LIMIT 10000
437 last unless scalar(@$r);
438 while (defined($n = shift @$r)) {
439 for my $i ($cur..($n - 1)) {
440 my $idx = idx_shard($self, $i);
441 $idx->ipc_do('xdb_remove_quiet', $i);
442 $active_shards{$idx} = $idx;
446 if (checkpoint_due($sync)) {
447 for my $idx (values %active_shards) {
448 $nr += $idx->ipc_do('nr_quiet_rm')
451 reindex_checkpoint($self, $sync);
454 warn "I: eliminated $nr stale Xapian documents\n" if $nr != 0;
458 my ($self, $opt) = @_;
459 $self->{cfg} or die "E: GC requires ->attach_config\n";
462 need_checkpoint => \(my $need_checkpoint = 0),
464 next_check => now() + 10,
465 checkpoint_unlocks => 1,
469 $self->idx_init($opt); # acquire lock via V2Writable::_idx_init
470 eidx_gc_scan_inboxes($self, $sync);
471 eidx_gc_scan_shards($self, $sync);
476 my ($self, $sync, $smsg) = @_;
477 my $ibx_id = delete($smsg->{ibx_id}) // die '{ibx_id} unset';
478 my $pos = $sync->{id2pos}->{$ibx_id} // die "$ibx_id no pos";
479 $self->{-ibx_ary_known}->[$pos] //
480 die "BUG: ibx for $smsg->{blob} not mapped"
483 sub _fd_constrained ($) {
485 $self->{-fd_constrained} //= do {
487 if (eval { require BSD::Resource; 1 }) {
488 my $NOFILE = BSD::Resource::RLIMIT_NOFILE();
489 ($soft, undef) = BSD::Resource::getrlimit($NOFILE);
491 chomp($soft = `sh -c 'ulimit -n'`);
493 if (defined($soft)) {
494 # $want is an estimate
495 my $want = scalar(@{$self->{ibx_active}}) + 64;
496 my $ret = $want > $soft;
499 RLIMIT_NOFILE=$soft insufficient (want: $want), will close DB handles early
504 warn "Unable to determine RLIMIT_NOFILE: $@\n";
510 sub _reindex_finalize ($$$) {
511 my ($req, $smsg, $eml) = @_;
512 my $sync = $req->{sync};
513 my $self = $sync->{self};
514 my $by_chash = delete $req->{by_chash} or die 'BUG: no {by_chash}';
515 my $nr = scalar(keys(%$by_chash)) or die 'BUG: no content hashes';
516 my $orig_smsg = $req->{orig_smsg} // die 'BUG: no {orig_smsg}';
517 my $docid = $smsg->{num} = $orig_smsg->{num};
518 $self->{oidx}->add_overview($eml, $smsg); # may rethread
519 check_batch_limit({ %$sync, new_smsg => $smsg });
520 my $chash0 = $smsg->{chash} // die "BUG: $smsg->{blob} no {chash}";
521 my $stable = delete($by_chash->{$chash0}) //
522 die "BUG: $smsg->{blob} chash missing";
523 my $idx = $self->idx_shard($docid);
524 my $top_smsg = pop @$stable;
525 $top_smsg == $smsg or die 'BUG: top_smsg != smsg';
526 my $ibx = _ibx_for($self, $sync, $smsg);
527 $idx->index_eml($eml, $smsg, $ibx->eidx_key);
528 for my $x (reverse @$stable) {
529 $ibx = _ibx_for($self, $sync, $x);
530 my $hdr = delete $x->{hdr} // die 'BUG: no {hdr}';
531 $idx->ipc_do('add_eidx_info', $docid, $ibx->eidx_key, $hdr);
533 return if $nr == 1; # likely, all good
535 warn "W: #$docid split into $nr due to deduplication change\n";
537 for my $ary (values %$by_chash) {
538 for my $x (reverse @$ary) {
539 warn "removing #$docid xref3 $x->{blob}\n";
540 my $n = $self->{oidx}->remove_xref3($docid, $x->{blob});
541 die "BUG: $x->{blob} invalidated #$docid" if $n == 0;
543 my $x = pop(@$ary) // die "BUG: #$docid {by_chash} empty";
544 $x->{num} = delete($x->{xnum}) // die '{xnum} unset';
545 $ibx = _ibx_for($self, $sync, $x);
546 if (my $over = $ibx->over) {
547 my $e = $over->get_art($x->{num});
548 $e->{blob} eq $x->{blob} or die <<EOF;
549 $x->{blob} != $e->{blob} (${\$ibx->eidx_key}:$e->{num});
551 push @todo, $ibx, $e;
552 $over->dbh_close if _fd_constrained($self);
554 die "$ibx->{inboxdir}: over.sqlite3 unusable: $!\n";
558 while (my ($ibx, $e) = splice(@todo, 0, 2)) {
559 reindex_unseen($self, $sync, $ibx, $e);
563 sub _reindex_oid { # git->cat_async callback
564 my ($bref, $oid, $type, $size, $req) = @_;
565 my $sync = $req->{sync};
566 my $self = $sync->{self};
567 my $orig_smsg = $req->{orig_smsg} // die 'BUG: no {orig_smsg}';
568 my $expect_oid = $req->{xr3r}->[$req->{ix}]->[2];
569 my $docid = $orig_smsg->{num};
570 if (is_bad_blob($oid, $type, $size, $expect_oid)) {
571 my $remain = $self->{oidx}->remove_xref3($docid, $expect_oid);
573 warn "W: #$docid gone or corrupted\n";
574 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
575 } elsif (my $next_oid = $req->{xr3r}->[++$req->{ix}]->[2]) {
576 $self->git->cat_async($next_oid, \&_reindex_oid, $req);
578 warn "BUG: #$docid gone (UNEXPECTED)\n";
579 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
583 my $ci = $self->{current_info};
584 local $self->{current_info} = "$ci #$docid $oid";
585 my $re_smsg = bless { blob => $oid }, 'PublicInbox::Smsg';
586 $re_smsg->set_bytes($$bref, $size);
587 my $eml = PublicInbox::Eml->new($bref);
588 $re_smsg->populate($eml, { autime => $orig_smsg->{ds},
589 cotime => $orig_smsg->{ts} });
590 my $chash = content_hash($eml);
591 $re_smsg->{chash} = $chash;
592 $re_smsg->{xnum} = $req->{xr3r}->[$req->{ix}]->[1];
593 $re_smsg->{ibx_id} = $req->{xr3r}->[$req->{ix}]->[0];
594 $re_smsg->{hdr} = $eml->header_obj;
595 push @{$req->{by_chash}->{$chash}}, $re_smsg;
596 if (my $next_oid = $req->{xr3r}->[++$req->{ix}]->[2]) {
597 $self->git->cat_async($next_oid, \&_reindex_oid, $req);
598 } else { # last $re_smsg is the highest priority xref3
599 local $self->{current_info} = "$ci #$docid";
600 _reindex_finalize($req, $re_smsg, $eml);
604 sub _reindex_smsg ($$$) {
605 my ($self, $sync, $smsg) = @_;
606 my $docid = $smsg->{num};
607 my $xr3 = $self->{oidx}->get_xref3($docid, 1);
608 if (scalar(@$xr3) == 0) { # _reindex_check_stale should've covered this
610 BUG? #$docid $smsg->{blob} is not referenced by inboxes during reindex
612 $self->{oidx}->delete_by_num($docid);
613 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
617 # we sort {xr3r} in the reverse order of ibx_sorted so we can
618 # hit the common case in _reindex_finalize without rereading
619 # from git (or holding multiple messages in memory).
620 my $id2pos = $sync->{id2pos}; # index in ibx_sorted
622 $id2pos->{$b->[0]} <=> $id2pos->{$a->[0]}
624 $b->[1] <=> $a->[1] # break ties with {xnum}
626 @$xr3 = map { [ $_->[0], $_->[1], unpack('H*', $_->[2]) ] } @$xr3;
627 my $req = { orig_smsg => $smsg, sync => $sync, xr3r => $xr3, ix => 0 };
628 $self->git->cat_async($xr3->[$req->{ix}]->[2], \&_reindex_oid, $req);
631 sub checkpoint_due ($) {
633 ${$sync->{need_checkpoint}} || (now() > $sync->{next_check});
637 # I've copied FS images and only changed the hostname before,
638 # so prepend hostname. Use `state' since these a BOFH can change
639 # these while this process is running and we always want to be
640 # able to release locks taken by this process.
641 state $retval = hostname . '-' . do {
642 my $m; # machine-id(5) is systemd
643 if (open(my $fh, '<', '/etc/machine-id')) { $m = <$fh> }
644 # (g)hostid(1) is in GNU coreutils, kern.hostid is most BSDs
645 chomp($m ||= `{ sysctl -n kern.hostid ||
646 hostid || ghostid; } 2>/dev/null`
647 || "no-machine-id-or-hostid-on-$^O");
654 my $expect = delete($self->{-eidxq_locked}) or return;
655 my ($owner_pid, undef) = split(/-/, $expect);
656 return if $owner_pid != $$; # shards may fork
657 my $oidx = $self->{oidx};
659 my $cur = $oidx->eidx_meta('eidxq_lock') // '';
660 if ($cur eq $expect) {
661 $oidx->eidx_meta('eidxq_lock', '');
663 } elsif ($cur ne '') {
664 warn "E: eidxq_lock($expect) stolen by $cur\n";
666 warn "E: eidxq_lock($expect) released by another process\n";
673 eidxq_release($self) and $self->{oidx}->commit_lazy;
676 sub _eidxq_take ($) {
678 my $val = "$$-${\time}-$>-".host_ident;
679 $self->{oidx}->eidx_meta('eidxq_lock', $val);
680 $self->{-eidxq_locked} = $val;
683 sub eidxq_lock_acquire ($) {
685 my $oidx = $self->{oidx};
687 my $cur = $oidx->eidx_meta('eidxq_lock') || return _eidxq_take($self);
688 if (my $locked = $self->{-eidxq_locked}) { # be lazy
689 return $locked if $locked eq $cur;
691 my ($pid, $time, $euid, $ident) = split(/-/, $cur, 4);
692 my $t = strftime('%Y-%m-%d %k:%M:%S', gmtime($time));
693 if ($euid == $> && $ident eq host_ident) {
696 I: PID:$pid (re)indexing Xapian since $t, it will continue our work
700 warn "I: eidxq_lock is stale ($cur), clobbering\n";
701 return _eidxq_take($self);
703 warn "E: kill(0, $pid) failed: $!\n"; # fall-through:
705 my $fn = $oidx->dbh->sqlite_db_filename;
707 W: PID:$pid, UID:$euid on $ident is indexing Xapian since $t
708 W: If this is unexpected, delete `eidxq_lock' from the `eidx_meta' table:
709 W: sqlite3 $fn 'DELETE FROM eidx_meta WHERE key = "eidxq_lock"'
714 sub ibx_sorted ($$) {
715 my ($self, $type) = @_;
716 $self->{"-ibx_ary_$type"} //= do {
717 # highest boost first, stable for config-ordering tiebreaker
720 ($b->{boost} // 0) <=> ($a->{boost} // 0)
721 } @{$self->{'ibx_'.$type} // die "BUG: $type unknown"} ];
725 sub prep_id2pos ($) {
729 $id2pos{$_->{-ibx_id}} = $pos++ for (@{ibx_sorted($self, 'known')});
733 sub eidxq_process ($$) { # for reindexing
734 my ($self, $sync) = @_;
735 return unless $self->{cfg};
737 return unless eidxq_lock_acquire($self);
738 my $dbh = $self->{oidx}->dbh;
739 my $tot = $dbh->selectrow_array('SELECT COUNT(*) FROM eidxq') or return;
741 local $sync->{-regen_fmt} = "%u/$tot\n";
742 my $pr = $sync->{-opt}->{-progress};
744 my $min = $dbh->selectrow_array('SELECT MIN(docid) FROM eidxq');
745 my $max = $dbh->selectrow_array('SELECT MAX(docid) FROM eidxq');
746 $pr->("Xapian indexing $min..$max (total=$tot)\n");
748 $sync->{id2pos} //= prep_id2pos($self);
751 $del = $dbh->prepare('DELETE FROM eidxq WHERE docid = ?');
752 $iter = $dbh->prepare('SELECT docid FROM eidxq ORDER BY docid ASC');
754 while (defined(my $docid = $iter->fetchrow_array)) {
755 last if $sync->{quit};
756 if (my $smsg = $self->{oidx}->get_art($docid)) {
757 _reindex_smsg($self, $sync, $smsg);
759 warn "E: #$docid does not exist in over\n";
761 $del->execute($docid);
764 if (checkpoint_due($sync)) {
765 $dbh = $del = $iter = undef;
766 reindex_checkpoint($self, $sync); # release lock
767 $dbh = $self->{oidx}->dbh;
771 $self->git->async_wait_all;
772 $pr->("reindexed ${$sync->{nr}}/$tot\n") if $pr;
775 sub _reindex_unseen { # git->cat_async callback
776 my ($bref, $oid, $type, $size, $req) = @_;
777 return if is_bad_blob($oid, $type, $size, $req->{oid});
778 my $self = $req->{self} // die 'BUG: {self} unset';
779 local $self->{current_info} = "$self->{current_info} $oid";
780 my $new_smsg = bless { blob => $oid, }, 'PublicInbox::Smsg';
781 $new_smsg->set_bytes($$bref, $size);
782 my $eml = $req->{eml} = PublicInbox::Eml->new($bref);
783 $req->{new_smsg} = $new_smsg;
784 $req->{chash} = content_hash($eml);
785 $req->{mids} = mids($eml); # do_step iterates through this
786 do_step($req); # enter the normal indexing flow
789 # --reindex may catch totally unseen messages, this handles them
790 sub reindex_unseen ($$$$) {
791 my ($self, $sync, $ibx, $xsmsg) = @_;
794 autime => $xsmsg->{ds},
795 cotime => $xsmsg->{ts},
796 oid => $xsmsg->{blob},
798 xnum => $xsmsg->{num},
799 # {mids} and {chash} will be filled in at _reindex_unseen
801 warn "I: reindex_unseen ${\$ibx->eidx_key}:$req->{xnum}:$req->{oid}\n";
802 $self->git->cat_async($xsmsg->{blob}, \&_reindex_unseen, $req);
805 sub _unref_stale_range ($$$) {
806 my ($sync, $ibx, $lt_or_gt) = @_;
810 $r = $sync->{self}->{oidx}->dbh->selectall_arrayref(
811 <<EOS, undef, $ibx->{-ibx_id});
812 SELECT docid,xnum,oidbin FROM xref3
813 WHERE ibx_id = ? AND xnum $lt_or_gt LIMIT $lim
815 return if $sync->{quit};
816 for (@$r) { # hopefully rare, not worth optimizing:
817 my ($docid, $xnum, $oidbin) = @$_;
818 my $hex = unpack('H*', $oidbin);
819 warn("# $xnum:$hex (#$docid): stale\n");
820 _unref_doc($sync, $docid, $ibx, $xnum, $oidbin);
822 } while (scalar(@$r) == $lim);
826 sub _reindex_check_ibx ($$$) {
827 my ($self, $sync, $ibx) = @_;
828 my $ibx_id = $ibx->{-ibx_id};
830 my $opt = { limit => $slice };
831 my ($beg, $end) = (1, $slice);
832 my $err = sync_inbox($self, $sync, $ibx) and return;
833 my $max = $ibx->over->max;
834 $end = $max if $end > $max;
836 # first, check if we missed any messages in target $ibx
838 my $pr = $sync->{-opt}->{-progress};
839 my $ekey = $ibx->eidx_key;
840 local $sync->{-regen_fmt} = "$ekey checking %u/$max\n";
842 my $fast = $sync->{-opt}->{fast};
843 my $dsu; # _unref_stale_range (< $lo) called
845 while (scalar(@{$msgs = $ibx->over->query_xover($beg, $end, $opt)})) {
846 ${$sync->{nr}} = $beg;
847 $beg = $msgs->[-1]->{num} + 1;
848 $end = $beg + $slice;
849 $end = $max if $end > $max;
850 if (checkpoint_due($sync)) {
851 reindex_checkpoint($self, $sync); # release lock
853 ($lo, $hi) = ($msgs->[0]->{num}, $msgs->[-1]->{num});
854 $dsu //= _unref_stale_range($sync, $ibx, "< $lo");
855 my $x3a = $self->{oidx}->dbh->selectall_arrayref(
856 <<"", undef, $ibx_id, $lo, $hi);
857 SELECT xnum,oidbin,docid FROM xref3 WHERE
858 ibx_id = ? AND xnum >= ? AND xnum <= ?
862 my $k = pack('J', $_->[0]) . $_->[1];
863 push @{$x3m{$k}}, $_->[2];
866 for my $xsmsg (@$msgs) {
867 my $k = pack('JH*', $xsmsg->{num}, $xsmsg->{blob});
868 my $docids = delete($x3m{$k});
869 if (!defined($docids)) {
870 reindex_unseen($self, $sync, $ibx, $xsmsg);
872 for my $num (@$docids) {
873 $self->{oidx}->eidxq_add($num);
875 return if $sync->{quit};
878 return if $sync->{quit};
879 next unless scalar keys %x3m;
881 # eliminate stale/mismatched entries
882 my %mismatch = map { $_->{num} => $_->{blob} } @$msgs;
883 while (my ($k, $docids) = each %x3m) {
884 my ($xnum, $hex) = unpack('JH*', $k);
885 my $bin = pack('H*', $hex);
886 my $exp = $mismatch{$xnum};
887 my $m = defined($exp) ? "mismatch (!= $exp)" : 'stale';
888 warn("# $xnum:$hex (#@$docids): $m\n");
889 for my $i (@$docids) {
890 _unref_doc($sync, $i, $ibx, $xnum, $bin);
894 _unref_stale_range($sync, $ibx, "> $hi") if defined($hi);
897 sub _reindex_inbox ($$$) {
898 my ($self, $sync, $ibx) = @_;
899 my $ekey = $ibx->eidx_key;
900 local $self->{current_info} = $ekey;
901 if (defined(my $err = _ibx_index_reject($ibx))) {
902 warn "W: cannot reindex $ekey ($err)\n";
904 _reindex_check_ibx($self, $sync, $ibx);
906 delete @$ibx{qw(over mm search git)}; # won't need these for a bit
910 my ($self, $sync) = @_;
911 return unless $self->{cfg};
913 # acquire eidxq_lock early because full reindex takes forever
914 # and incremental -extindex processes can run during our checkpoints
915 if (!eidxq_lock_acquire($self)) {
916 warn "E: aborting --reindex\n";
919 for my $ibx (@{ibx_sorted($self, 'active')}) {
920 _reindex_inbox($self, $sync, $ibx);
921 last if $sync->{quit};
923 $self->git->async_wait_all; # ensure eidxq gets filled completely
924 eidxq_process($self, $sync) unless $sync->{quit};
928 my ($self, $sync, $ibx) = @_;
929 my $err = _sync_inbox($self, $sync, $ibx);
930 delete @$ibx{qw(mm over)};
931 warn $err, "\n" if defined($err);
935 sub dd_smsg { # git->cat_async callback
936 my ($bref, $oid, $type, $size, $dd) = @_;
937 my $smsg = $dd->{smsg} // die 'BUG: dd->{smsg} missing';
938 my $self = $dd->{self} // die 'BUG: {self} missing';
939 my $per_mid = $dd->{per_mid} // die 'BUG: {per_mid} missing';
940 if ($type eq 'missing') {
941 _blob_missing($dd, $smsg);
942 } elsif (!is_bad_blob($oid, $type, $size, $smsg->{blob})) {
943 local $self->{current_info} = "$self->{current_info} $oid";
944 my $chash = content_hash(PublicInbox::Eml->new($bref));
945 push(@{$per_mid->{dd_chash}->{$chash}}, $smsg);
947 return if $per_mid->{last_smsg} != $smsg;
948 while (my ($chash, $ary) = each %{$per_mid->{dd_chash}}) {
949 my $keep = shift @$ary;
950 next if !scalar(@$ary);
951 $per_mid->{sync}->{dedupe_cull} += scalar(@$ary);
953 "# <$keep->{mid}> keeping #$keep->{num}, dropping ",
954 join(', ', map { "#$_->{num}" } @$ary),"\n";
955 next if $per_mid->{sync}->{-opt}->{'dry-run'};
956 my $oidx = $self->{oidx};
957 for my $smsg (@$ary) {
958 my $gone = $smsg->{num};
959 $oidx->merge_xref3($keep->{num}, $gone, $smsg->{blob});
960 $self->idx_shard($gone)->ipc_do('xdb_remove', $gone);
961 $oidx->delete_by_num($gone);
966 sub eidx_dedupe ($$$) {
967 my ($self, $sync, $msgids) = @_;
968 $sync->{dedupe_cull} = 0;
971 return unless eidxq_lock_acquire($self);
972 my ($iter, $cur_mid);
975 my ($max_id) = $self->{oidx}->dbh->selectrow_array(<<EOS);
976 SELECT MAX(id) FROM msgid
978 local $sync->{-regen_fmt} = "dedupe %u/$max_id\n";
980 # note: we could write this query more intelligently,
981 # but that causes lock contention with read-only processes
983 $cur_mid = $msgids->[$idx];
984 if ($cur_mid eq '') { # all Message-IDs
985 $iter = $self->{oidx}->dbh->prepare(<<EOS);
986 SELECT mid,id FROM msgid WHERE id > ? ORDER BY id ASC
988 $iter->execute($min_id);
990 $iter = $self->{oidx}->dbh->prepare(<<EOS);
991 SELECT mid,id FROM msgid WHERE mid = ? AND id > ? ORDER BY id ASC
993 $iter->execute($cur_mid, $min_id);
995 while (my ($mid, $id) = $iter->fetchrow_array) {
996 last if $sync->{quit};
997 $self->{current_info} = "dedupe $mid";
998 ${$sync->{nr}} = $min_id = $id;
1000 while (my $x = $self->{oidx}->next_by_mid($mid, \$id, \$prv)) {
1003 next if scalar(@smsg) < 2;
1005 dd_chash => {}, # chash => [ary of smsgs]
1006 last_smsg => $smsg[-1],
1010 $candidates += scalar(@smsg) - 1;
1011 for my $smsg (@smsg) {
1013 per_mid => $per_mid,
1017 $self->git->cat_async($smsg->{blob}, \&dd_smsg, $dd);
1019 # need to wait on every single one @smsg contents can get
1020 # invalidated inside dd_smsg for messages with multiple
1022 $self->git->async_wait_all;
1024 if (checkpoint_due($sync)) {
1026 reindex_checkpoint($self, $sync);
1027 goto dedupe_restart;
1030 goto dedupe_restart if defined($msgids->[++$idx]);
1032 my $n = delete $sync->{dedupe_cull};
1033 if (my $pr = $sync->{-opt}->{-progress}) {
1034 $pr->("culled $n/$candidates candidates ($nr_mid msgids)\n");
1039 sub eidx_sync { # main entry point
1040 my ($self, $opt) = @_;
1042 my $warn_cb = $SIG{__WARN__} || \&CORE::warn;
1043 local $self->{current_info} = '';
1044 local $SIG{__WARN__} = sub {
1045 return if PublicInbox::Eml::warn_ignore(@_);
1046 $warn_cb->($self->{current_info}, ': ', @_);
1048 $self->idx_init($opt); # acquire lock via V2Writable::_idx_init
1049 $self->{oidx}->rethread_prepare($opt);
1051 need_checkpoint => \(my $need_checkpoint = 0),
1053 next_check => now() + 10,
1055 # DO NOT SET {reindex} here, it's incompatible with reused
1056 # V2Writable code, reindex is totally different here
1057 # compared to v1/v2 inboxes because we have multiple histories
1059 -regen_fmt => "%u/?\n",
1061 local $SIG{USR1} = sub { $need_checkpoint = 1 };
1062 my $quit = PublicInbox::SearchIdx::quit_cb($sync);
1063 local $SIG{QUIT} = $quit;
1064 local $SIG{INT} = $quit;
1065 local $SIG{TERM} = $quit;
1066 for my $ibx (@{ibx_sorted($self, 'known')}) {
1067 $ibx->{-ibx_id} //= $self->{oidx}->ibx_id($ibx->eidx_key);
1070 if (scalar(grep { defined($_->{boost}) } @{$self->{ibx_known}})) {
1071 $sync->{id2pos} //= prep_id2pos($self);
1072 $sync->{boost_in_use} = 1;
1075 if (my $msgids = delete($opt->{dedupe})) {
1076 local $sync->{checkpoint_unlocks} = 1;
1077 eidx_dedupe($self, $sync, $msgids);
1079 if (delete($opt->{reindex})) {
1080 local $sync->{checkpoint_unlocks} = 1;
1081 eidx_reindex($self, $sync);
1084 # don't use $_ here, it'll get clobbered by reindex_checkpoint
1085 if ($opt->{scan} // 1) {
1086 for my $ibx (@{ibx_sorted($self, 'active')}) {
1087 last if $sync->{quit};
1088 sync_inbox($self, $sync, $ibx);
1091 $self->{oidx}->rethread_done($opt) unless $sync->{quit};
1092 eidxq_process($self, $sync) unless $sync->{quit};
1094 eidxq_release($self);
1096 $sync; # for eidx_watch
1099 sub update_last_commit { # overrides V2Writable
1100 my ($self, $sync, $stk) = @_;
1101 my $unit = $sync->{unit} // return;
1102 my $latest_cmt = $stk ? $stk->{latest_cmt} : ${$sync->{latest_cmt}};
1103 defined($latest_cmt) or return;
1104 my $ibx = $sync->{ibx} or die 'BUG: {ibx} missing';
1105 my $ekey = $ibx->eidx_key;
1106 my $uv = $ibx->uidvalidity;
1107 my $epoch = $unit->{epoch};
1109 my $v = $ibx->version;
1111 die 'No {epoch} for v2 unit' unless defined $epoch;
1112 $meta_key = "lc-v2:$ekey//$uv;$epoch";
1114 die 'Unexpected {epoch} for v1 unit' if defined $epoch;
1115 $meta_key = "lc-v1:$ekey//$uv";
1117 die "Unsupported inbox version: $v";
1119 my $last = $self->{oidx}->eidx_meta($meta_key);
1120 if (defined $last && is_ancestor($self->git, $last, $latest_cmt)) {
1121 my @cmd = (qw(rev-list --count), "$last..$latest_cmt");
1122 chomp(my $n = $unit->{git}->qx(@cmd));
1123 return if $n ne '' && $n == 0;
1125 $self->{oidx}->eidx_meta($meta_key, $latest_cmt);
1128 sub _idx_init { # with_umask callback
1129 my ($self, $opt) = @_;
1130 PublicInbox::V2Writable::_idx_init($self, $opt); # acquires ei.lock
1131 $self->{midx} = PublicInbox::MiscIdx->new($self);
1134 sub symlink_packs ($$) {
1135 my ($ibx, $pd) = @_;
1137 my $glob = "$ibx->{inboxdir}/git/*.git/objects/pack/*.idx";
1138 for my $idx (bsd_glob($glob, GLOB_NOSORT)) {
1139 my $src = substr($idx, 0, -length('.idx'));
1140 my $dst = $pd . substr($src, rindex($src, '/'));
1141 if (-f "$src.pack" and
1142 symlink("$src.pack", "$dst.pack") and
1143 symlink($idx, "$dst.idx") and
1146 # .promisor, .bitmap, .rev and .keep are optional
1147 # XXX should we symlink .keep here?
1148 for my $s (qw(promisor bitmap rev)) {
1149 symlink("$src.$s", "$dst.$s") if -f "$src.$s";
1151 } elsif (!$!{EEXIST}) {
1152 warn "W: ln -s $src.{pack,idx} => $dst.*: $!\n";
1153 unlink "$dst.pack", "$dst.idx";
1159 sub idx_init { # similar to V2Writable
1160 my ($self, $opt) = @_;
1161 return if $self->{idx_shards};
1163 $self->git->cleanup;
1165 my $ALL = $self->git->{git_dir}; # topdir/ALL.git
1166 my ($has_new, $alt, $seen);
1167 if ($opt->{-private}) { # LeiStore
1168 my $local = "$self->{topdir}/local"; # lei/store
1169 $self->{mg} //= PublicInbox::MultiGit->new($self->{topdir},
1170 'ALL.git', 'local');
1173 umask 077; # don't bother restoring for lei
1174 PublicInbox::Import::init_bare($ALL);
1175 $self->git->qx(qw(config core.sharedRepository 0600));
1177 ($alt, $seen) = $self->{mg}->read_alternates(\$mode);
1178 $has_new = $self->{mg}->merge_epochs($alt, $seen);
1179 } else { # extindex has no epochs
1180 $self->{mg} //= PublicInbox::MultiGit->new($self->{topdir},
1182 ($alt, $seen) = $self->{mg}->read_alternates(\$mode,
1184 PublicInbox::Import::init_bare($ALL);
1187 # git-multi-pack-index(1) can speed up "git cat-file" startup slightly
1189 my $pd = "$ALL/objects/pack";
1190 if (opendir(my $dh, $pd)) { # 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;
1197 } elsif ($!{ENOENT}) {
1198 mkdir($pd) or die "mkdir($pd): $!";
1200 die "opendir($pd): $!";
1203 for my $ibx (@{ibx_sorted($self, 'active')}) {
1204 # create symlinks for multi-pack-index
1205 $git_midx += symlink_packs($ibx, $pd);
1206 # add new lines to our alternates file
1207 my $d = $ibx->git->{git_dir} . '/objects';
1208 next if exists $alt->{$d};
1209 if (my @st = stat($d)) {
1210 next if $seen->{"$st[0]\0$st[1]"}++;
1212 warn "W: stat($d) failed (from $ibx->{inboxdir}): $!\n";
1213 next if $opt->{-idx_gc};
1217 ($has_new || $new ne '') and
1218 $self->{mg}->write_alternates($mode, $alt, $new);
1219 $git_midx and $self->with_umask(sub {
1220 my @cmd = ('multi-pack-index');
1221 push @cmd, '--no-progress' if ($opt->{quiet}//0) > 1;
1222 my $lk = $self->lock_for_scope;
1223 system('git', "--git-dir=$ALL", @cmd, 'write');
1224 # ignore errors, fairly new command, may not exist
1226 $self->parallel_init($self->{indexlevel});
1227 $self->with_umask(\&_idx_init, $self, $opt);
1228 $self->{oidx}->begin_lazy;
1229 $self->{oidx}->eidx_prep;
1230 $self->{midx}->create_xdb if $new ne '';
1233 sub _watch_commit { # PublicInbox::DS::add_timer callback
1235 delete $self->{-commit_timer};
1236 eidxq_process($self, $self->{-watch_sync});
1237 eidxq_release($self);
1238 my $fmt = delete $self->{-watch_sync}->{-regen_fmt};
1239 reindex_checkpoint($self, $self->{-watch_sync});
1240 $self->{-watch_sync}->{-regen_fmt} = $fmt;
1242 # call event_step => done unless commit_timer is armed
1243 PublicInbox::DS::requeue($self);
1246 sub on_inbox_unlock { # called by PublicInbox::InboxIdle
1247 my ($self, $ibx) = @_;
1248 my $opt = $self->{-watch_sync}->{-opt};
1249 my $pr = $opt->{-progress};
1250 my $ekey = $ibx->eidx_key;
1251 local $0 = "sync $ekey";
1252 $pr->("indexing $ekey\n") if $pr;
1253 $self->idx_init($opt);
1254 sync_inbox($self, $self->{-watch_sync}, $ibx);
1255 $self->{-commit_timer} //= add_timer($opt->{'commit-interval'} // 10,
1256 \&_watch_commit, $self);
1259 sub eidx_reload { # -extindex --watch SIGHUP handler
1260 my ($self, $idler) = @_;
1262 my $pr = $self->{-watch_sync}->{-opt}->{-progress};
1263 $pr->('reloading ...') if $pr;
1264 delete $self->{-resync_queue};
1265 delete $self->{-ibx_ary_known};
1266 delete $self->{-ibx_ary_active};
1267 $self->{ibx_known} = [];
1268 $self->{ibx_active} = [];
1269 %{$self->{ibx_map}} = ();
1270 delete $self->{-watch_sync}->{id2pos};
1271 my $cfg = PublicInbox::Config->new;
1272 attach_config($self, $cfg);
1273 $idler->refresh($cfg);
1274 $pr->(" done\n") if $pr;
1276 warn "reload not supported without --all\n";
1280 sub eidx_resync_start ($) { # -extindex --watch SIGUSR1 handler
1282 $self->{-resync_queue} //= [ @{ibx_sorted($self, 'active')} ];
1283 PublicInbox::DS::requeue($self); # trigger our ->event_step
1286 sub event_step { # PublicInbox::DS::requeue callback
1288 if (my $resync_queue = $self->{-resync_queue}) {
1289 if (my $ibx = shift(@$resync_queue)) {
1290 on_inbox_unlock($self, $ibx);
1291 PublicInbox::DS::requeue($self);
1293 delete $self->{-resync_queue};
1294 _watch_commit($self);
1297 done($self) unless $self->{-commit_timer};
1301 sub eidx_watch { # public-inbox-extindex --watch main loop
1302 my ($self, $opt) = @_;
1303 local @SIG{keys %SIG} = values %SIG;
1304 for my $sig (qw(HUP USR1 TSTP QUIT INT TERM)) {
1305 $SIG{$sig} = sub { warn "SIG$sig ignored while scanning\n" };
1307 require PublicInbox::InboxIdle;
1308 require PublicInbox::DS;
1309 require PublicInbox::Syscall;
1310 require PublicInbox::Sigfd;
1311 my $idler = PublicInbox::InboxIdle->new($self->{cfg});
1312 if (!$self->{cfg}) {
1313 $idler->watch_inbox($_) for (@{ibx_sorted($self, 'active')});
1315 for my $ibx (@{ibx_sorted($self, 'active')}) {
1316 $ibx->subscribe_unlock(__PACKAGE__, $self)
1318 my $pr = $opt->{-progress};
1319 $pr->("performing initial scan ...\n") if $pr;
1320 my $sync = eidx_sync($self, $opt); # initial sync
1321 return if $sync->{quit};
1322 my $oldset = PublicInbox::DS::block_signals();
1323 local $self->{current_info} = '';
1324 my $cb = $SIG{__WARN__} || \&CORE::warn;
1325 local $SIG{__WARN__} = sub {
1326 return if PublicInbox::Eml::warn_ignore(@_);
1327 $cb->($self->{current_info}, ': ', @_);
1330 HUP => sub { eidx_reload($self, $idler) },
1331 USR1 => sub { eidx_resync_start($self) },
1332 TSTP => sub { kill('STOP', $$) },
1334 my $quit = PublicInbox::SearchIdx::quit_cb($sync);
1335 $sig->{QUIT} = $sig->{INT} = $sig->{TERM} = $quit;
1336 local $self->{-watch_sync} = $sync; # for ->on_inbox_unlock
1337 PublicInbox::DS->SetPostLoopCallback(sub { !$sync->{quit} });
1338 $pr->("initial scan complete, entering event loop\n") if $pr;
1339 # calls InboxIdle->event_step:
1340 PublicInbox::DS::event_loop($sig, $oldset);
1345 *done = \&PublicInbox::V2Writable::done;
1346 *with_umask = \&PublicInbox::InboxWritable::with_umask;
1347 *parallel_init = \&PublicInbox::V2Writable::parallel_init;
1348 *nproc_shards = \&PublicInbox::V2Writable::nproc_shards;
1349 *sync_prepare = \&PublicInbox::V2Writable::sync_prepare;
1350 *index_todo = \&PublicInbox::V2Writable::index_todo;
1351 *count_shards = \&PublicInbox::V2Writable::count_shards;
1352 *atfork_child = \&PublicInbox::V2Writable::atfork_child;
1353 *idx_shard = \&PublicInbox::V2Writable::idx_shard;
1354 *reindex_checkpoint = \&PublicInbox::V2Writable::reindex_checkpoint;
1355 *checkpoint = \&PublicInbox::V2Writable::checkpoint;