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);
133 my ($req, $smsg) = @_;
134 my $self = $req->{self};
135 my $docid = $smsg->{num};
136 my $idx = $self->idx_shard($docid);
137 my $oid = $req->{oid};
138 my $xibx = $req->{ibx};
139 my $eml = $req->{eml};
140 my $eidx_key = $xibx->eidx_key;
141 if (my $new_smsg = $req->{new_smsg}) { # 'm' on cross-posted message
142 my $xnum = $req->{xnum};
143 $self->{oidx}->add_xref3($docid, $xnum, $oid, $eidx_key);
144 $idx->ipc_do('add_eidx_info', $docid, $eidx_key, $eml);
145 apply_boost($req, $smsg) if $req->{boost_in_use};
148 my $nr = $self->{oidx}->remove_xref3($docid, $oid, $eidx_key,
151 $self->{oidx}->eidxq_del($docid);
152 $idx->ipc_do('xdb_remove', $docid);
153 } elsif ($rm_eidx_info) {
154 $idx->ipc_do('remove_eidx_info',
155 $docid, $eidx_key, $eml);
156 $self->{oidx}->eidxq_add($docid); # yes, add
161 # called by V2Writable::sync_prepare
162 sub artnum_max { $_[0]->{oidx}->eidx_max }
164 sub index_unseen ($) {
166 my $new_smsg = $req->{new_smsg} or die 'BUG: {new_smsg} unset';
167 my $eml = delete $req->{eml};
168 $new_smsg->populate($eml, $req);
169 my $self = $req->{self};
170 my $docid = $self->{oidx}->adj_counter('eidx_docid', '+');
171 $new_smsg->{num} = $docid;
172 my $idx = $self->idx_shard($docid);
173 $self->{oidx}->add_overview($eml, $new_smsg);
174 my $oid = $new_smsg->{blob};
175 my $ibx = delete $req->{ibx} or die 'BUG: {ibx} unset';
176 $self->{oidx}->add_xref3($docid, $req->{xnum}, $oid, $ibx->eidx_key);
177 $idx->index_eml($eml, $new_smsg, $ibx->eidx_key);
178 check_batch_limit($req);
181 sub do_finalize ($) {
183 if (my $indexed = $req->{indexed}) { # duplicated messages
184 do_xpost($req, $_) for @$indexed;
185 } elsif (exists $req->{new_smsg}) { # totally unseen messsage
188 # `d' message was already unindexed in the v1/v2 inboxes,
189 # so it's too noisy to warn, here.
191 # cur_cmt may be undef for unindex_oid, set by V2Writable::index_todo
192 if (defined(my $cur_cmt = $req->{cur_cmt})) {
193 ${$req->{latest_cmt}} = $cur_cmt;
197 sub do_step ($) { # main iterator for adding messages to the index
199 my $self = $req->{self} // die 'BUG: {self} missing';
201 if (my $next_arg = $req->{next_arg}) {
202 if (my $smsg = $self->{oidx}->next_by_mid(@$next_arg)) {
203 $req->{cur_smsg} = $smsg;
204 $self->git->cat_async($smsg->{blob},
205 \&ck_existing, $req);
206 return; # ck_existing calls do_step
208 delete $req->{next_arg};
210 die "BUG: {cur_smsg} still set" if $req->{cur_smsg};
211 my $mid = shift(@{$req->{mids}}) // last;
213 $req->{next_arg} = [ $mid, \$id, \$prev ];
219 sub _blob_missing ($$) { # called when $smsg->{blob} is bad
220 my ($req, $smsg) = @_;
221 my $self = $req->{self};
222 my $xref3 = $self->{oidx}->get_xref3($smsg->{num});
223 my @keep = grep(!/:$smsg->{blob}\z/, @$xref3);
225 warn "E: $smsg->{blob} gone, removing #$smsg->{num}\n";
226 $keep[0] =~ /:([a-f0-9]{40,}+)\z/ or
227 die "BUG: xref $keep[0] has no OID";
229 $self->{oidx}->remove_xref3($smsg->{num}, $smsg->{blob});
230 $self->{oidx}->update_blob($smsg, $oidhex) or warn <<EOM;
231 E: #$smsg->{num} gone ($smsg->{blob} => $oidhex)
234 warn "E: $smsg->{blob} gone, removing #$smsg->{num}\n";
235 $self->{oidx}->delete_by_num($smsg->{num});
239 sub ck_existing { # git->cat_async callback
240 my ($bref, $oid, $type, $size, $req) = @_;
241 my $smsg = delete $req->{cur_smsg} or die 'BUG: {cur_smsg} missing';
242 if ($type eq 'missing') {
243 _blob_missing($req, $smsg);
244 } elsif (!is_bad_blob($oid, $type, $size, $smsg->{blob})) {
245 my $self = $req->{self} // die 'BUG: {self} missing';
246 local $self->{current_info} = "$self->{current_info} $oid";
247 my $cur = PublicInbox::Eml->new($bref);
248 if (content_hash($cur) eq $req->{chash}) {
249 push @{$req->{indexed}}, $smsg; # for do_xpost
250 } # else { index_unseen later }
255 # is the messages visible in the inbox currently being indexed?
256 # return the number if so
257 sub cur_ibx_xnum ($$) {
258 my ($req, $bref) = @_;
259 my $ibx = $req->{ibx} or die 'BUG: current {ibx} missing';
261 $req->{eml} = PublicInbox::Eml->new($bref);
262 $req->{chash} = content_hash($req->{eml});
263 $req->{mids} = mids($req->{eml});
264 for my $mid (@{$req->{mids}}) {
266 while (my $x = $ibx->over->next_by_mid($mid, \$id, \$prev)) {
267 return $x->{num} if $x->{blob} eq $req->{oid};
273 sub index_oid { # git->cat_async callback for 'm'
274 my ($bref, $oid, $type, $size, $req) = @_;
275 my $self = $req->{self};
276 local $self->{current_info} = "$self->{current_info} $oid";
277 return if is_bad_blob($oid, $type, $size, $req->{oid});
278 my $new_smsg = $req->{new_smsg} = bless {
280 }, 'PublicInbox::Smsg';
281 $new_smsg->set_bytes($$bref, $size);
282 defined($req->{xnum} = cur_ibx_xnum($req, $bref)) or return;
287 sub unindex_oid { # git->cat_async callback for 'd'
288 my ($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, $req->{oid});
292 return if defined(cur_ibx_xnum($req, $bref)); # was re-added
296 # overrides V2Writable::last_commits, called by sync_ranges via sync_prepare
298 my ($self, $sync) = @_;
300 my $ekey = $sync->{ibx}->eidx_key;
301 my $uv = $sync->{ibx}->uidvalidity;
302 for my $i (0..$sync->{epoch_max}) {
303 $heads->[$i] = $self->{oidx}->eidx_meta("lc-v2:$ekey//$uv;$i");
308 sub _ibx_index_reject ($) {
310 $ibx->mm // return 'unindexed, no msgmap.sqlite3';
311 $ibx->uidvalidity // return 'no UIDVALIDITY';
312 $ibx->over // return 'unindexed, no over.sqlite3';
316 sub _sync_inbox ($$$) {
317 my ($self, $sync, $ibx) = @_;
318 my $ekey = $ibx->eidx_key;
319 if (defined(my $err = _ibx_index_reject($ibx))) {
320 return "W: skipping $ekey ($err)";
323 $sync->{nr} = \(my $nr = 0);
324 my $v = $ibx->version;
326 $sync->{epoch_max} = $ibx->max_git_epoch // return;
327 sync_prepare($self, $sync); # or return # TODO: once MiscIdx is stable
329 my $uv = $ibx->uidvalidity;
330 my $lc = $self->{oidx}->eidx_meta("lc-v1:$ekey//$uv");
331 my $head = $ibx->mm->last_commit //
332 return "E: $ibx->{inboxdir} is not indexed";
333 my $stk = prepare_stack($sync, $lc ? "$lc..$head" : $head);
334 my $unit = { stack => $stk, git => $ibx->git };
335 push @{$sync->{todo}}, $unit;
337 return "E: $ekey unsupported inbox version (v$v)";
339 for my $unit (@{delete($sync->{todo}) // []}) {
340 last if $sync->{quit};
341 index_todo($self, $sync, $unit);
343 $self->{midx}->index_ibx($ibx) unless $sync->{quit};
344 $ibx->git->cleanup; # done with this inbox, now
348 sub gc_unref_doc ($$$$) {
349 my ($self, $ibx_id, $eidx_key, $docid) = @_;
351 # for debug/info purposes, oids may no longer be accessible
352 my $dbh = $self->{oidx}->dbh;
353 my $sth = $dbh->prepare_cached(<<'', undef, 1);
354 SELECT oidbin FROM xref3 WHERE docid = ? AND ibx_id = ?
356 $sth->execute($docid, $ibx_id);
357 my @oid = map { unpack('H*', $_->[0]) } @{$sth->fetchall_arrayref};
359 $remain += $self->{oidx}->remove_xref3($docid, $oid, $eidx_key);
362 $self->{oidx}->eidxq_add($docid); # enqueue for reindex
364 warn "I: unref #$docid $eidx_key $oid\n";
367 warn "I: remove #$docid $eidx_key @oid\n";
368 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
372 sub eidx_gc_scan_inboxes ($$) {
373 my ($self, $sync) = @_;
374 my ($x3_doc, $ibx_ck);
376 $x3_doc = $self->{oidx}->dbh->prepare(<<EOM);
377 SELECT docid FROM xref3 WHERE ibx_id = ?
379 $ibx_ck = $self->{oidx}->dbh->prepare(<<EOM);
380 SELECT ibx_id,eidx_key FROM inboxes
383 while (my ($ibx_id, $eidx_key) = $ibx_ck->fetchrow_array) {
384 next if $self->{ibx_map}->{$eidx_key};
385 $self->{midx}->remove_eidx_key($eidx_key);
386 warn "I: deleting messages for $eidx_key...\n";
387 $x3_doc->execute($ibx_id);
388 while (defined(my $docid = $x3_doc->fetchrow_array)) {
389 gc_unref_doc($self, $ibx_id, $eidx_key, $docid);
390 if (checkpoint_due($sync)) {
391 $x3_doc = $ibx_ck = undef;
392 reindex_checkpoint($self, $sync);
396 $self->{oidx}->dbh->do(<<'', undef, $ibx_id);
397 DELETE FROM inboxes WHERE ibx_id = ?
399 # drop last_commit info
401 $pat =~ s/([_%\\])/\\$1/g;
402 $self->{oidx}->dbh->do('PRAGMA case_sensitive_like = ON');
403 my $lc_i = $self->{oidx}->dbh->prepare(<<'');
404 SELECT key FROM eidx_meta WHERE key LIKE ? ESCAPE ?
406 $lc_i->execute("lc-%:$pat//%", '\\');
407 while (my ($key) = $lc_i->fetchrow_array) {
408 next if $key !~ m!\Alc-v[1-9]+:\Q$eidx_key\E//!;
409 warn "I: removing $key\n";
410 $self->{oidx}->dbh->do(<<'', undef, $key);
411 DELETE FROM eidx_meta WHERE key = ?
414 warn "I: $eidx_key removed\n";
418 sub eidx_gc_scan_shards ($$) { # TODO: use for lei/store
419 my ($self, $sync) = @_;
420 my $nr = $self->{oidx}->dbh->do(<<'');
421 DELETE FROM xref3 WHERE docid NOT IN (SELECT num FROM over)
423 warn "I: eliminated $nr stale xref3 entries\n" if $nr != 0;
425 # fixup from old bugs:
426 $nr = $self->{oidx}->dbh->do(<<'');
427 DELETE FROM over WHERE num NOT IN (SELECT docid FROM xref3)
429 warn "I: eliminated $nr stale over entries\n" if $nr != 0;
431 my ($cur) = $self->{oidx}->dbh->selectrow_array(<<EOM);
432 SELECT MIN(num) FROM over
434 my ($max) = $self->{oidx}->dbh->selectrow_array(<<EOM);
435 SELECT MAX(num) FROM over
439 $exists = $self->{oidx}->dbh->prepare(<<EOM);
440 SELECT COUNT(num) FROM over WHERE num = ?
442 for (; $cur <= $max; $cur++) {
443 $exists->execute($cur);
444 next if $exists->fetchrow_array != 0;
445 $self->idx_shard($cur)->ipc_do('xdb_remove_quiet', $cur);
446 if (checkpoint_due($sync)) {
448 reindex_checkpoint($self, $sync);
455 my ($self, $opt) = @_;
456 $self->{cfg} or die "E: GC requires ->attach_config\n";
459 need_checkpoint => \(my $need_checkpoint = 0),
461 next_check => now() + 10,
462 checkpoint_unlocks => 1,
465 $self->idx_init($opt); # acquire lock via V2Writable::_idx_init
466 eidx_gc_scan_inboxes($self, $sync);
467 eidx_gc_scan_shards($self, $sync);
472 my ($self, $sync, $smsg) = @_;
473 my $ibx_id = delete($smsg->{ibx_id}) // die '{ibx_id} unset';
474 my $pos = $sync->{id2pos}->{$ibx_id} // die "$ibx_id no pos";
475 $self->{-ibx_ary_known}->[$pos] //
476 die "BUG: ibx for $smsg->{blob} not mapped"
479 sub _fd_constrained ($) {
481 $self->{-fd_constrained} //= do {
483 if (eval { require BSD::Resource; 1 }) {
484 my $NOFILE = BSD::Resource::RLIMIT_NOFILE();
485 ($soft, undef) = BSD::Resource::getrlimit($NOFILE);
487 chomp($soft = `sh -c 'ulimit -n'`);
489 if (defined($soft)) {
490 # $want is an estimate
491 my $want = scalar(@{$self->{ibx_active}}) + 64;
492 my $ret = $want > $soft;
495 RLIMIT_NOFILE=$soft insufficient (want: $want), will close DB handles early
500 warn "Unable to determine RLIMIT_NOFILE: $@\n";
506 sub _reindex_finalize ($$$) {
507 my ($req, $smsg, $eml) = @_;
508 my $sync = $req->{sync};
509 my $self = $sync->{self};
510 my $by_chash = delete $req->{by_chash} or die 'BUG: no {by_chash}';
511 my $nr = scalar(keys(%$by_chash)) or die 'BUG: no content hashes';
512 my $orig_smsg = $req->{orig_smsg} // die 'BUG: no {orig_smsg}';
513 my $docid = $smsg->{num} = $orig_smsg->{num};
514 $self->{oidx}->add_overview($eml, $smsg); # may rethread
515 check_batch_limit({ %$sync, new_smsg => $smsg });
516 my $chash0 = $smsg->{chash} // die "BUG: $smsg->{blob} no {chash}";
517 my $stable = delete($by_chash->{$chash0}) //
518 die "BUG: $smsg->{blob} chash missing";
519 my $idx = $self->idx_shard($docid);
520 my $top_smsg = pop @$stable;
521 $top_smsg == $smsg or die 'BUG: top_smsg != smsg';
522 my $ibx = _ibx_for($self, $sync, $smsg);
523 $idx->index_eml($eml, $smsg, $ibx->eidx_key);
524 for my $x (reverse @$stable) {
525 $ibx = _ibx_for($self, $sync, $x);
526 my $hdr = delete $x->{hdr} // die 'BUG: no {hdr}';
527 $idx->ipc_do('add_eidx_info', $docid, $ibx->eidx_key, $hdr);
529 return if $nr == 1; # likely, all good
531 warn "W: #$docid split into $nr due to deduplication change\n";
533 for my $ary (values %$by_chash) {
534 for my $x (reverse @$ary) {
535 warn "removing #$docid xref3 $x->{blob}\n";
536 my $n = $self->{oidx}->remove_xref3($docid, $x->{blob});
537 die "BUG: $x->{blob} invalidated #$docid" if $n == 0;
539 my $x = pop(@$ary) // die "BUG: #$docid {by_chash} empty";
540 $x->{num} = delete($x->{xnum}) // die '{xnum} unset';
541 $ibx = _ibx_for($self, $sync, $x);
542 if (my $over = $ibx->over) {
543 my $e = $over->get_art($x->{num});
544 $e->{blob} eq $x->{blob} or die <<EOF;
545 $x->{blob} != $e->{blob} (${\$ibx->eidx_key}:$e->{num});
547 push @todo, $ibx, $e;
548 $over->dbh_close if _fd_constrained($self);
550 die "$ibx->{inboxdir}: over.sqlite3 unusable: $!\n";
554 while (my ($ibx, $e) = splice(@todo, 0, 2)) {
555 reindex_unseen($self, $sync, $ibx, $e);
559 sub _reindex_oid { # git->cat_async callback
560 my ($bref, $oid, $type, $size, $req) = @_;
561 my $sync = $req->{sync};
562 my $self = $sync->{self};
563 my $orig_smsg = $req->{orig_smsg} // die 'BUG: no {orig_smsg}';
564 my $expect_oid = $req->{xr3r}->[$req->{ix}]->[2];
565 my $docid = $orig_smsg->{num};
566 if (is_bad_blob($oid, $type, $size, $expect_oid)) {
567 my $remain = $self->{oidx}->remove_xref3($docid, $expect_oid);
569 warn "W: #$docid gone or corrupted\n";
570 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
571 } elsif (my $next_oid = $req->{xr3r}->[++$req->{ix}]->[2]) {
572 $self->git->cat_async($next_oid, \&_reindex_oid, $req);
574 warn "BUG: #$docid gone (UNEXPECTED)\n";
575 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
579 my $ci = $self->{current_info};
580 local $self->{current_info} = "$ci #$docid $oid";
581 my $re_smsg = bless { blob => $oid }, 'PublicInbox::Smsg';
582 $re_smsg->set_bytes($$bref, $size);
583 my $eml = PublicInbox::Eml->new($bref);
584 $re_smsg->populate($eml, { autime => $orig_smsg->{ds},
585 cotime => $orig_smsg->{ts} });
586 my $chash = content_hash($eml);
587 $re_smsg->{chash} = $chash;
588 $re_smsg->{xnum} = $req->{xr3r}->[$req->{ix}]->[1];
589 $re_smsg->{ibx_id} = $req->{xr3r}->[$req->{ix}]->[0];
590 $re_smsg->{hdr} = $eml->header_obj;
591 push @{$req->{by_chash}->{$chash}}, $re_smsg;
592 if (my $next_oid = $req->{xr3r}->[++$req->{ix}]->[2]) {
593 $self->git->cat_async($next_oid, \&_reindex_oid, $req);
594 } else { # last $re_smsg is the highest priority xref3
595 local $self->{current_info} = "$ci #$docid";
596 _reindex_finalize($req, $re_smsg, $eml);
600 sub _reindex_smsg ($$$) {
601 my ($self, $sync, $smsg) = @_;
602 my $docid = $smsg->{num};
603 my $xr3 = $self->{oidx}->get_xref3($docid, 1);
604 if (scalar(@$xr3) == 0) { # _reindex_check_stale should've covered this
606 BUG? #$docid $smsg->{blob} is not referenced by inboxes during reindex
608 $self->{oidx}->delete_by_num($docid);
609 $self->idx_shard($docid)->ipc_do('xdb_remove', $docid);
613 # we sort {xr3r} in the reverse order of ibx_sorted so we can
614 # hit the common case in _reindex_finalize without rereading
615 # from git (or holding multiple messages in memory).
616 my $id2pos = $sync->{id2pos}; # index in ibx_sorted
618 $id2pos->{$b->[0]} <=> $id2pos->{$a->[0]}
620 $b->[1] <=> $a->[1] # break ties with {xnum}
622 @$xr3 = map { [ $_->[0], $_->[1], unpack('H*', $_->[2]) ] } @$xr3;
623 my $req = { orig_smsg => $smsg, sync => $sync, xr3r => $xr3, ix => 0 };
624 $self->git->cat_async($xr3->[$req->{ix}]->[2], \&_reindex_oid, $req);
627 sub checkpoint_due ($) {
629 ${$sync->{need_checkpoint}} || (now() > $sync->{next_check});
633 # I've copied FS images and only changed the hostname before,
634 # so prepend hostname. Use `state' since these a BOFH can change
635 # these while this process is running and we always want to be
636 # able to release locks taken by this process.
637 state $retval = hostname . '-' . do {
638 my $m; # machine-id(5) is systemd
639 if (open(my $fh, '<', '/etc/machine-id')) { $m = <$fh> }
640 # (g)hostid(1) is in GNU coreutils, kern.hostid is most BSDs
641 chomp($m ||= `{ sysctl -n kern.hostid ||
642 hostid || ghostid; } 2>/dev/null`
643 || "no-machine-id-or-hostid-on-$^O");
650 my $expect = delete($self->{-eidxq_locked}) or return;
651 my ($owner_pid, undef) = split(/-/, $expect);
652 return if $owner_pid != $$; # shards may fork
653 my $oidx = $self->{oidx};
655 my $cur = $oidx->eidx_meta('eidxq_lock') // '';
656 if ($cur eq $expect) {
657 $oidx->eidx_meta('eidxq_lock', '');
659 } elsif ($cur ne '') {
660 warn "E: eidxq_lock($expect) stolen by $cur\n";
662 warn "E: eidxq_lock($expect) released by another process\n";
669 eidxq_release($self) and $self->{oidx}->commit_lazy;
672 sub _eidxq_take ($) {
674 my $val = "$$-${\time}-$>-".host_ident;
675 $self->{oidx}->eidx_meta('eidxq_lock', $val);
676 $self->{-eidxq_locked} = $val;
679 sub eidxq_lock_acquire ($) {
681 my $oidx = $self->{oidx};
683 my $cur = $oidx->eidx_meta('eidxq_lock') || return _eidxq_take($self);
684 if (my $locked = $self->{-eidxq_locked}) { # be lazy
685 return $locked if $locked eq $cur;
687 my ($pid, $time, $euid, $ident) = split(/-/, $cur, 4);
688 my $t = strftime('%Y-%m-%d %k:%M:%S', gmtime($time));
689 if ($euid == $> && $ident eq host_ident) {
692 I: PID:$pid (re)indexing Xapian since $t, it will continue our work
696 warn "I: eidxq_lock is stale ($cur), clobbering\n";
697 return _eidxq_take($self);
699 warn "E: kill(0, $pid) failed: $!\n"; # fall-through:
701 my $fn = $oidx->dbh->sqlite_db_filename;
703 W: PID:$pid, UID:$euid on $ident is indexing Xapian since $t
704 W: If this is unexpected, delete `eidxq_lock' from the `eidx_meta' table:
705 W: sqlite3 $fn 'DELETE FROM eidx_meta WHERE key = "eidxq_lock"'
710 sub ibx_sorted ($$) {
711 my ($self, $type) = @_;
712 $self->{"-ibx_ary_$type"} //= do {
713 # highest boost first, stable for config-ordering tiebreaker
716 ($b->{boost} // 0) <=> ($a->{boost} // 0)
717 } @{$self->{'ibx_'.$type} // die "BUG: $type unknown"} ];
721 sub prep_id2pos ($) {
725 $id2pos{$_->{-ibx_id}} = $pos++ for (@{ibx_sorted($self, 'known')});
729 sub eidxq_process ($$) { # for reindexing
730 my ($self, $sync) = @_;
731 return unless $self->{cfg};
733 return unless eidxq_lock_acquire($self);
734 my $dbh = $self->{oidx}->dbh;
735 my $tot = $dbh->selectrow_array('SELECT COUNT(*) FROM eidxq') or return;
737 local $sync->{-regen_fmt} = "%u/$tot\n";
738 my $pr = $sync->{-opt}->{-progress};
740 my $min = $dbh->selectrow_array('SELECT MIN(docid) FROM eidxq');
741 my $max = $dbh->selectrow_array('SELECT MAX(docid) FROM eidxq');
742 $pr->("Xapian indexing $min..$max (total=$tot)\n");
744 $sync->{id2pos} //= prep_id2pos($self);
747 $del = $dbh->prepare('DELETE FROM eidxq WHERE docid = ?');
748 $iter = $dbh->prepare('SELECT docid FROM eidxq ORDER BY docid ASC');
750 while (defined(my $docid = $iter->fetchrow_array)) {
751 last if $sync->{quit};
752 if (my $smsg = $self->{oidx}->get_art($docid)) {
753 _reindex_smsg($self, $sync, $smsg);
755 warn "E: #$docid does not exist in over\n";
757 $del->execute($docid);
760 if (checkpoint_due($sync)) {
761 $dbh = $del = $iter = undef;
762 reindex_checkpoint($self, $sync); # release lock
763 $dbh = $self->{oidx}->dbh;
767 $self->git->async_wait_all;
768 $pr->("reindexed ${$sync->{nr}}/$tot\n") if $pr;
771 sub _reindex_unseen { # git->cat_async callback
772 my ($bref, $oid, $type, $size, $req) = @_;
773 return if is_bad_blob($oid, $type, $size, $req->{oid});
774 my $self = $req->{self} // die 'BUG: {self} unset';
775 local $self->{current_info} = "$self->{current_info} $oid";
776 my $new_smsg = bless { blob => $oid, }, 'PublicInbox::Smsg';
777 $new_smsg->set_bytes($$bref, $size);
778 my $eml = $req->{eml} = PublicInbox::Eml->new($bref);
779 $req->{new_smsg} = $new_smsg;
780 $req->{chash} = content_hash($eml);
781 $req->{mids} = mids($eml); # do_step iterates through this
782 do_step($req); # enter the normal indexing flow
785 # --reindex may catch totally unseen messages, this handles them
786 sub reindex_unseen ($$$$) {
787 my ($self, $sync, $ibx, $xsmsg) = @_;
790 autime => $xsmsg->{ds},
791 cotime => $xsmsg->{ts},
792 oid => $xsmsg->{blob},
794 xnum => $xsmsg->{num},
795 # {mids} and {chash} will be filled in at _reindex_unseen
797 warn "I: reindex_unseen ${\$ibx->eidx_key}:$req->{xnum}:$req->{oid}\n";
798 $self->git->cat_async($xsmsg->{blob}, \&_reindex_unseen, $req);
801 sub _reindex_check_unseen ($$$) {
802 my ($self, $sync, $ibx) = @_;
803 my $ibx_id = $ibx->{-ibx_id};
805 my ($beg, $end) = (1, $slice);
807 # first, check if we missed any messages in target $ibx
809 my $pr = $sync->{-opt}->{-progress};
810 my $ekey = $ibx->eidx_key;
811 local $sync->{-regen_fmt} =
812 "$ekey checking unseen %u/".$ibx->over->max."\n";
814 my $fast = $sync->{-opt}->{fast};
815 while (scalar(@{$msgs = $ibx->over->query_xover($beg, $end)})) {
816 ${$sync->{nr}} = $beg;
817 $beg = $msgs->[-1]->{num} + 1;
818 $end = $beg + $slice;
819 if (checkpoint_due($sync)) {
820 reindex_checkpoint($self, $sync); # release lock
823 my $inx3 = $self->{oidx}->dbh->prepare_cached(<<'', undef, 1);
824 SELECT DISTINCT(docid) FROM xref3 WHERE
825 ibx_id = ? AND xnum = ? AND oidbin = ?
827 for my $xsmsg (@$msgs) {
828 my $oidbin = pack('H*', $xsmsg->{blob});
829 $inx3->bind_param(1, $ibx_id);
830 $inx3->bind_param(2, $xsmsg->{num});
831 $inx3->bind_param(3, $oidbin, SQL_BLOB);
833 my $docids = $inx3->fetchall_arrayref;
834 # index messages which were totally missed
835 # the first time around ASAP:
836 if (scalar(@$docids) == 0) {
837 reindex_unseen($self, $sync, $ibx, $xsmsg);
838 } elsif (!$fast) { # already seen, reindex later
839 for my $r (@$docids) {
840 $self->{oidx}->eidxq_add($r->[0]);
843 last if $sync->{quit};
845 last if $sync->{quit};
849 sub _reindex_check_stale ($$$) {
850 my ($self, $sync, $ibx) = @_;
852 my $pr = $sync->{-opt}->{-progress};
854 my $ekey = $ibx->eidx_key;
855 local $sync->{-regen_fmt} =
856 "$ekey checking stale/missing %u/".$ibx->over->max."\n";
859 if (checkpoint_due($sync)) {
860 reindex_checkpoint($self, $sync); # release lock
862 # now, check if there's stale xrefs
863 my $iter = $self->{oidx}->dbh->prepare_cached(<<'', undef, 1);
864 SELECT docid,xnum,oidbin FROM xref3 WHERE ibx_id = ? AND docid > ?
865 ORDER BY docid,xnum ASC LIMIT 10000
867 $iter->execute($ibx->{-ibx_id}, $min);
870 while (my ($docid, $xnum, $oidbin) = $iter->fetchrow_array) {
871 return if $sync->{quit};
872 ${$sync->{nr}} = $xnum;
874 $fetching = $min = $docid;
875 my $smsg = $ibx->over->get_art($xnum);
879 } elsif (pack('H*', $smsg->{blob}) ne $oidbin) {
880 $err = "mismatch (!= $smsg->{blob})";
882 next; # likely, all good
884 # current_info already has eidx_key
885 my $oidhex = unpack('H*', $oidbin);
886 warn "$xnum:$oidhex (#$docid): $err\n";
887 my $del = $self->{oidx}->dbh->prepare_cached(<<'');
888 DELETE FROM xref3 WHERE ibx_id = ? AND xnum = ? AND oidbin = ?
890 $del->bind_param(1, $ibx->{-ibx_id});
891 $del->bind_param(2, $xnum);
892 $del->bind_param(3, $oidbin, SQL_BLOB);
895 # get_xref3 over-fetches, but this is a rare path:
896 my $xr3 = $self->{oidx}->get_xref3($docid);
897 my $idx = $self->idx_shard($docid);
898 if (scalar(@$xr3) == 0) { # all gone
899 $self->{oidx}->delete_by_num($docid);
900 $self->{oidx}->eidxq_del($docid);
901 $idx->ipc_do('xdb_remove', $docid);
902 } else { # enqueue for reindex of remaining messages
903 $idx->ipc_do('remove_eidx_info',
904 $docid, $ibx->eidx_key);
905 $self->{oidx}->eidxq_add($docid); # yes, add
908 } while (defined $fetching);
911 sub _reindex_inbox ($$$) {
912 my ($self, $sync, $ibx) = @_;
913 my $ekey = $ibx->eidx_key;
914 local $self->{current_info} = $ekey;
915 if (defined(my $err = _ibx_index_reject($ibx))) {
916 warn "W: cannot reindex $ekey ($err)\n";
918 _reindex_check_unseen($self, $sync, $ibx);
919 _reindex_check_stale($self, $sync, $ibx) unless $sync->{quit};
921 delete @$ibx{qw(over mm search git)}; # won't need these for a bit
925 my ($self, $sync) = @_;
926 return unless $self->{cfg};
928 # acquire eidxq_lock early because full reindex takes forever
929 # and incremental -extindex processes can run during our checkpoints
930 if (!eidxq_lock_acquire($self)) {
931 warn "E: aborting --reindex\n";
934 for my $ibx (@{ibx_sorted($self, 'active')}) {
935 _reindex_inbox($self, $sync, $ibx);
936 last if $sync->{quit};
938 $self->git->async_wait_all; # ensure eidxq gets filled completely
939 eidxq_process($self, $sync) unless $sync->{quit};
943 my ($self, $sync, $ibx) = @_;
944 my $err = _sync_inbox($self, $sync, $ibx);
945 delete @$ibx{qw(mm over)};
946 warn $err, "\n" if defined($err);
949 sub dd_smsg { # git->cat_async callback
950 my ($bref, $oid, $type, $size, $dd) = @_;
951 my $smsg = $dd->{smsg} // die 'BUG: dd->{smsg} missing';
952 my $self = $dd->{self} // die 'BUG: {self} missing';
953 my $per_mid = $dd->{per_mid} // die 'BUG: {per_mid} missing';
954 if ($type eq 'missing') {
955 _blob_missing($dd, $smsg);
956 } elsif (!is_bad_blob($oid, $type, $size, $smsg->{blob})) {
957 local $self->{current_info} = "$self->{current_info} $oid";
958 my $chash = content_hash(PublicInbox::Eml->new($bref));
959 push(@{$per_mid->{dd_chash}->{$chash}}, $smsg);
961 return if $per_mid->{last_smsg} != $smsg;
962 while (my ($chash, $ary) = each %{$per_mid->{dd_chash}}) {
963 my $keep = shift @$ary;
964 next if !scalar(@$ary);
965 $per_mid->{sync}->{dedupe_cull} += scalar(@$ary);
967 "# <$keep->{mid}> keeping #$keep->{num}, dropping ",
968 join(', ', map { "#$_->{num}" } @$ary),"\n";
969 next if $per_mid->{sync}->{-opt}->{'dry-run'};
970 my $oidx = $self->{oidx};
971 for my $smsg (@$ary) {
972 my $gone = $smsg->{num};
973 $oidx->merge_xref3($keep->{num}, $gone, $smsg->{blob});
974 $self->idx_shard($gone)->ipc_do('xdb_remove', $gone);
975 $oidx->delete_by_num($gone);
980 sub eidx_dedupe ($$$) {
981 my ($self, $sync, $msgids) = @_;
982 $sync->{dedupe_cull} = 0;
985 return unless eidxq_lock_acquire($self);
986 my ($iter, $cur_mid);
989 my ($max_id) = $self->{oidx}->dbh->selectrow_array(<<EOS);
990 SELECT MAX(id) FROM msgid
992 local $sync->{-regen_fmt} = "dedupe %u/$max_id\n";
994 # note: we could write this query more intelligently,
995 # but that causes lock contention with read-only processes
997 $cur_mid = $msgids->[$idx];
998 if ($cur_mid eq '') { # all Message-IDs
999 $iter = $self->{oidx}->dbh->prepare(<<EOS);
1000 SELECT mid,id FROM msgid WHERE id > ? ORDER BY id ASC
1002 $iter->execute($min_id);
1004 $iter = $self->{oidx}->dbh->prepare(<<EOS);
1005 SELECT mid,id FROM msgid WHERE mid = ? AND id > ? ORDER BY id ASC
1007 $iter->execute($cur_mid, $min_id);
1009 while (my ($mid, $id) = $iter->fetchrow_array) {
1010 last if $sync->{quit};
1011 $self->{current_info} = "dedupe $mid";
1012 ${$sync->{nr}} = $min_id = $id;
1014 while (my $x = $self->{oidx}->next_by_mid($mid, \$id, \$prv)) {
1017 next if scalar(@smsg) < 2;
1019 dd_chash => {}, # chash => [ary of smsgs]
1020 last_smsg => $smsg[-1],
1024 $candidates += scalar(@smsg) - 1;
1025 for my $smsg (@smsg) {
1027 per_mid => $per_mid,
1031 $self->git->cat_async($smsg->{blob}, \&dd_smsg, $dd);
1033 # need to wait on every single one @smsg contents can get
1034 # invalidated inside dd_smsg for messages with multiple
1036 $self->git->async_wait_all;
1038 if (checkpoint_due($sync)) {
1040 reindex_checkpoint($self, $sync);
1041 goto dedupe_restart;
1044 goto dedupe_restart if defined($msgids->[++$idx]);
1046 my $n = delete $sync->{dedupe_cull};
1047 if (my $pr = $sync->{-opt}->{-progress}) {
1048 $pr->("culled $n/$candidates candidates ($nr_mid msgids)\n");
1053 sub eidx_sync { # main entry point
1054 my ($self, $opt) = @_;
1056 my $warn_cb = $SIG{__WARN__} || \&CORE::warn;
1057 local $self->{current_info} = '';
1058 local $SIG{__WARN__} = sub {
1059 return if PublicInbox::Eml::warn_ignore(@_);
1060 $warn_cb->($self->{current_info}, ': ', @_);
1062 $self->idx_init($opt); # acquire lock via V2Writable::_idx_init
1063 $self->{oidx}->rethread_prepare($opt);
1065 need_checkpoint => \(my $need_checkpoint = 0),
1067 next_check => now() + 10,
1069 # DO NOT SET {reindex} here, it's incompatible with reused
1070 # V2Writable code, reindex is totally different here
1071 # compared to v1/v2 inboxes because we have multiple histories
1073 -regen_fmt => "%u/?\n",
1075 local $SIG{USR1} = sub { $need_checkpoint = 1 };
1076 my $quit = PublicInbox::SearchIdx::quit_cb($sync);
1077 local $SIG{QUIT} = $quit;
1078 local $SIG{INT} = $quit;
1079 local $SIG{TERM} = $quit;
1080 for my $ibx (@{ibx_sorted($self, 'known')}) {
1081 $ibx->{-ibx_id} //= $self->{oidx}->ibx_id($ibx->eidx_key);
1084 if (scalar(grep { defined($_->{boost}) } @{$self->{ibx_known}})) {
1085 $sync->{id2pos} //= prep_id2pos($self);
1086 $sync->{boost_in_use} = 1;
1089 if (my $msgids = delete($opt->{dedupe})) {
1090 local $sync->{checkpoint_unlocks} = 1;
1091 eidx_dedupe($self, $sync, $msgids);
1093 if (delete($opt->{reindex})) {
1094 local $sync->{checkpoint_unlocks} = 1;
1095 eidx_reindex($self, $sync);
1098 # don't use $_ here, it'll get clobbered by reindex_checkpoint
1099 if ($opt->{scan} // 1) {
1100 for my $ibx (@{ibx_sorted($self, 'active')}) {
1101 last if $sync->{quit};
1102 sync_inbox($self, $sync, $ibx);
1105 $self->{oidx}->rethread_done($opt) unless $sync->{quit};
1106 eidxq_process($self, $sync) unless $sync->{quit};
1108 eidxq_release($self);
1110 $sync; # for eidx_watch
1113 sub update_last_commit { # overrides V2Writable
1114 my ($self, $sync, $stk) = @_;
1115 my $unit = $sync->{unit} // return;
1116 my $latest_cmt = $stk ? $stk->{latest_cmt} : ${$sync->{latest_cmt}};
1117 defined($latest_cmt) or return;
1118 my $ibx = $sync->{ibx} or die 'BUG: {ibx} missing';
1119 my $ekey = $ibx->eidx_key;
1120 my $uv = $ibx->uidvalidity;
1121 my $epoch = $unit->{epoch};
1123 my $v = $ibx->version;
1125 die 'No {epoch} for v2 unit' unless defined $epoch;
1126 $meta_key = "lc-v2:$ekey//$uv;$epoch";
1128 die 'Unexpected {epoch} for v1 unit' if defined $epoch;
1129 $meta_key = "lc-v1:$ekey//$uv";
1131 die "Unsupported inbox version: $v";
1133 my $last = $self->{oidx}->eidx_meta($meta_key);
1134 if (defined $last && is_ancestor($self->git, $last, $latest_cmt)) {
1135 my @cmd = (qw(rev-list --count), "$last..$latest_cmt");
1136 chomp(my $n = $unit->{git}->qx(@cmd));
1137 return if $n ne '' && $n == 0;
1139 $self->{oidx}->eidx_meta($meta_key, $latest_cmt);
1142 sub _idx_init { # with_umask callback
1143 my ($self, $opt) = @_;
1144 PublicInbox::V2Writable::_idx_init($self, $opt); # acquires ei.lock
1145 $self->{midx} = PublicInbox::MiscIdx->new($self);
1148 sub symlink_packs ($$) {
1149 my ($ibx, $pd) = @_;
1151 my $glob = "$ibx->{inboxdir}/git/*.git/objects/pack/*.idx";
1152 for my $idx (bsd_glob($glob, GLOB_NOSORT)) {
1153 my $src = substr($idx, 0, -length('.idx'));
1154 my $dst = $pd . substr($src, rindex($src, '/'));
1155 if (-f "$src.pack" and
1156 symlink("$src.pack", "$dst.pack") and
1157 symlink($idx, "$dst.idx") and
1160 # .promisor, .bitmap, .rev and .keep are optional
1161 # XXX should we symlink .keep here?
1162 for my $s (qw(promisor bitmap rev)) {
1163 symlink("$src.$s", "$dst.$s") if -f "$src.$s";
1165 } elsif (!$!{EEXIST}) {
1166 warn "W: ln -s $src.{pack,idx} => $dst.*: $!\n";
1167 unlink "$dst.pack", "$dst.idx";
1173 sub idx_init { # similar to V2Writable
1174 my ($self, $opt) = @_;
1175 return if $self->{idx_shards};
1177 $self->git->cleanup;
1179 my $ALL = $self->git->{git_dir}; # topdir/ALL.git
1180 my ($has_new, $alt, $seen);
1181 if ($opt->{-private}) { # LeiStore
1182 my $local = "$self->{topdir}/local"; # lei/store
1183 $self->{mg} //= PublicInbox::MultiGit->new($self->{topdir},
1184 'ALL.git', 'local');
1187 umask 077; # don't bother restoring for lei
1188 PublicInbox::Import::init_bare($ALL);
1189 $self->git->qx(qw(config core.sharedRepository 0600));
1191 ($alt, $seen) = $self->{mg}->read_alternates(\$mode);
1192 $has_new = $self->{mg}->merge_epochs($alt, $seen);
1193 } else { # extindex has no epochs
1194 $self->{mg} //= PublicInbox::MultiGit->new($self->{topdir},
1196 ($alt, $seen) = $self->{mg}->read_alternates(\$mode,
1198 PublicInbox::Import::init_bare($ALL);
1201 # git-multi-pack-index(1) can speed up "git cat-file" startup slightly
1203 my $pd = "$ALL/objects/pack";
1204 if (opendir(my $dh, $pd)) { # drop stale symlinks
1205 while (defined(my $dn = readdir($dh))) {
1206 if ($dn =~ /\.(?:idx|pack|promisor|bitmap|rev)\z/) {
1208 unlink($f) if -l $f && !-e $f;
1211 } elsif ($!{ENOENT}) {
1212 mkdir($pd) or die "mkdir($pd): $!";
1214 die "opendir($pd): $!";
1217 for my $ibx (@{ibx_sorted($self, 'active')}) {
1218 # create symlinks for multi-pack-index
1219 $git_midx += symlink_packs($ibx, $pd);
1220 # add new lines to our alternates file
1221 my $d = $ibx->git->{git_dir} . '/objects';
1222 next if exists $alt->{$d};
1223 if (my @st = stat($d)) {
1224 next if $seen->{"$st[0]\0$st[1]"}++;
1226 warn "W: stat($d) failed (from $ibx->{inboxdir}): $!\n";
1227 next if $opt->{-idx_gc};
1231 ($has_new || $new ne '') and
1232 $self->{mg}->write_alternates($mode, $alt, $new);
1233 $git_midx and $self->with_umask(sub {
1234 my @cmd = ('multi-pack-index');
1235 push @cmd, '--no-progress' if ($opt->{quiet}//0) > 1;
1236 my $lk = $self->lock_for_scope;
1237 system('git', "--git-dir=$ALL", @cmd, 'write');
1238 # ignore errors, fairly new command, may not exist
1240 $self->parallel_init($self->{indexlevel});
1241 $self->with_umask(\&_idx_init, $self, $opt);
1242 $self->{oidx}->begin_lazy;
1243 $self->{oidx}->eidx_prep;
1244 $self->{midx}->create_xdb if $new ne '';
1247 sub _watch_commit { # PublicInbox::DS::add_timer callback
1249 delete $self->{-commit_timer};
1250 eidxq_process($self, $self->{-watch_sync});
1251 eidxq_release($self);
1252 my $fmt = delete $self->{-watch_sync}->{-regen_fmt};
1253 reindex_checkpoint($self, $self->{-watch_sync});
1254 $self->{-watch_sync}->{-regen_fmt} = $fmt;
1256 # call event_step => done unless commit_timer is armed
1257 PublicInbox::DS::requeue($self);
1260 sub on_inbox_unlock { # called by PublicInbox::InboxIdle
1261 my ($self, $ibx) = @_;
1262 my $opt = $self->{-watch_sync}->{-opt};
1263 my $pr = $opt->{-progress};
1264 my $ekey = $ibx->eidx_key;
1265 local $0 = "sync $ekey";
1266 $pr->("indexing $ekey\n") if $pr;
1267 $self->idx_init($opt);
1268 sync_inbox($self, $self->{-watch_sync}, $ibx);
1269 $self->{-commit_timer} //= add_timer($opt->{'commit-interval'} // 10,
1270 \&_watch_commit, $self);
1273 sub eidx_reload { # -extindex --watch SIGHUP handler
1274 my ($self, $idler) = @_;
1276 my $pr = $self->{-watch_sync}->{-opt}->{-progress};
1277 $pr->('reloading ...') if $pr;
1278 delete $self->{-resync_queue};
1279 delete $self->{-ibx_ary_known};
1280 delete $self->{-ibx_ary_active};
1281 $self->{ibx_known} = [];
1282 $self->{ibx_active} = [];
1283 %{$self->{ibx_map}} = ();
1284 delete $self->{-watch_sync}->{id2pos};
1285 my $cfg = PublicInbox::Config->new;
1286 attach_config($self, $cfg);
1287 $idler->refresh($cfg);
1288 $pr->(" done\n") if $pr;
1290 warn "reload not supported without --all\n";
1294 sub eidx_resync_start ($) { # -extindex --watch SIGUSR1 handler
1296 $self->{-resync_queue} //= [ @{ibx_sorted($self, 'active')} ];
1297 PublicInbox::DS::requeue($self); # trigger our ->event_step
1300 sub event_step { # PublicInbox::DS::requeue callback
1302 if (my $resync_queue = $self->{-resync_queue}) {
1303 if (my $ibx = shift(@$resync_queue)) {
1304 on_inbox_unlock($self, $ibx);
1305 PublicInbox::DS::requeue($self);
1307 delete $self->{-resync_queue};
1308 _watch_commit($self);
1311 done($self) unless $self->{-commit_timer};
1315 sub eidx_watch { # public-inbox-extindex --watch main loop
1316 my ($self, $opt) = @_;
1317 local @SIG{keys %SIG} = values %SIG;
1318 for my $sig (qw(HUP USR1 TSTP QUIT INT TERM)) {
1319 $SIG{$sig} = sub { warn "SIG$sig ignored while scanning\n" };
1321 require PublicInbox::InboxIdle;
1322 require PublicInbox::DS;
1323 require PublicInbox::Syscall;
1324 require PublicInbox::Sigfd;
1325 my $idler = PublicInbox::InboxIdle->new($self->{cfg});
1326 if (!$self->{cfg}) {
1327 $idler->watch_inbox($_) for (@{ibx_sorted($self, 'active')});
1329 for my $ibx (@{ibx_sorted($self, 'active')}) {
1330 $ibx->subscribe_unlock(__PACKAGE__, $self)
1332 my $pr = $opt->{-progress};
1333 $pr->("performing initial scan ...\n") if $pr;
1334 my $sync = eidx_sync($self, $opt); # initial sync
1335 return if $sync->{quit};
1336 my $oldset = PublicInbox::DS::block_signals();
1337 local $self->{current_info} = '';
1338 my $cb = $SIG{__WARN__} || \&CORE::warn;
1339 local $SIG{__WARN__} = sub {
1340 return if PublicInbox::Eml::warn_ignore(@_);
1341 $cb->($self->{current_info}, ': ', @_);
1344 HUP => sub { eidx_reload($self, $idler) },
1345 USR1 => sub { eidx_resync_start($self) },
1346 TSTP => sub { kill('STOP', $$) },
1348 my $quit = PublicInbox::SearchIdx::quit_cb($sync);
1349 $sig->{QUIT} = $sig->{INT} = $sig->{TERM} = $quit;
1350 local $self->{-watch_sync} = $sync; # for ->on_inbox_unlock
1351 PublicInbox::DS->SetPostLoopCallback(sub { !$sync->{quit} });
1352 $pr->("initial scan complete, entering event loop\n") if $pr;
1353 # calls InboxIdle->event_step:
1354 PublicInbox::DS::event_loop($sig, $oldset);
1359 *done = \&PublicInbox::V2Writable::done;
1360 *with_umask = \&PublicInbox::InboxWritable::with_umask;
1361 *parallel_init = \&PublicInbox::V2Writable::parallel_init;
1362 *nproc_shards = \&PublicInbox::V2Writable::nproc_shards;
1363 *sync_prepare = \&PublicInbox::V2Writable::sync_prepare;
1364 *index_todo = \&PublicInbox::V2Writable::index_todo;
1365 *count_shards = \&PublicInbox::V2Writable::count_shards;
1366 *atfork_child = \&PublicInbox::V2Writable::atfork_child;
1367 *idx_shard = \&PublicInbox::V2Writable::idx_shard;
1368 *reindex_checkpoint = \&PublicInbox::V2Writable::reindex_checkpoint;
1369 *checkpoint = \&PublicInbox::V2Writable::checkpoint;