user/dev discussion of public-inbox itself
 help / color / mirror / code / Atom feed
From: Eric Wong <e@yhbt.net>
To: meta@public-inbox.org
Subject: [PATCH 42/82] imap: start doing iterative config reloading
Date: Wed, 10 Jun 2020 07:04:39 +0000	[thread overview]
Message-ID: <20200610070519.18252-43-e@yhbt.net> (raw)
In-Reply-To: <20200610070519.18252-1-e@yhbt.net>

This will be used to prevent reloading a giant config with
tens/hundreds of thousands of inboxes from blocking the event
loop.
---
 lib/PublicInbox/Config.pm | 15 ++++++++++
 lib/PublicInbox/IMAPD.pm  | 61 +++++++++++++++++++++++++++++++++++----
 script/public-inbox-imapd |  2 +-
 3 files changed, 71 insertions(+), 7 deletions(-)

diff --git a/lib/PublicInbox/Config.pm b/lib/PublicInbox/Config.pm
index 458f29b2964..c18c9c75b4f 100644
--- a/lib/PublicInbox/Config.pm
+++ b/lib/PublicInbox/Config.pm
@@ -99,6 +99,21 @@ sub each_inbox {
 	}
 }
 
+sub iterate_start {
+	my ($self, $cb, $arg) = @_;
+	my $i = 0;
+	$self->{-iter} = [ \$i, $cb, $arg ];
+}
+
+# for PublicInbox::DS::next_tick
+sub event_step {
+	my ($self) = @_;
+	my ($i, $cb, $arg) = @{$self->{-iter}};
+	my $section = $self->{-section_order}->[$$i++];
+	delete($self->{-iter}) unless defined($section);
+	$cb->($self, $section, $arg);
+}
+
 sub lookup_newsgroup {
 	my ($self, $ng) = @_;
 	_lookup_fill($self, '-by_newsgroup', lc($ng));
diff --git a/lib/PublicInbox/IMAPD.pm b/lib/PublicInbox/IMAPD.pm
index 966879d8ea5..a10fabffb1f 100644
--- a/lib/PublicInbox/IMAPD.pm
+++ b/lib/PublicInbox/IMAPD.pm
@@ -7,6 +7,8 @@ package PublicInbox::IMAPD;
 use strict;
 use parent qw(PublicInbox::NNTPD);
 use PublicInbox::InboxIdle;
+use PublicInbox::IMAP;
+# *UID_BLOCK = \&PublicInbox::IMAP::UID_BLOCK;
 
 sub new {
 	my ($class) = @_;
@@ -51,17 +53,64 @@ sub refresh_inboxlist ($) {
 	$self->{inboxlist} = \@names;
 }
 
-sub refresh_groups {
-	my ($self) = @_;
-	my $pi_config = $self->{pi_config} = PublicInbox::Config->new;
-	$self->SUPER::refresh_groups($pi_config);
-	refresh_inboxlist($self);
+sub imapd_refresh_ibx { # pi_config->each_inbox cb
+	my ($ibx, $imapd) = @_;
+	my $ngname = $ibx->{newsgroup} or return;
+	if (ref $ngname) {
+		warn 'multiple newsgroups not supported: '.
+			join(', ', @$ngname). "\n";
+	} elsif ($ngname =~ m![^a-z0-9/_\.\-\~\@\+\=:]! ||
+		 $ngname =~ /\.[0-9]+-[0-9]+\z/) {
+		warn "mailbox name invalid: `$ngname'\n";
+	}
+
+	my $mm = $ibx->mm or return;
+	$ibx->{mm} = undef;
+	defined($ibx->{uidvalidity} = $mm->created_at) or return;
+	$imapd->{tmp_groups}->{$ngname} = $ibx;
 
-	if (my $idler = $self->{idler}) {
+	# preload to avoid fragmentation:
+	$ibx->description;
+	$ibx->base_url;
+	# my $max = $mm->max // 0;
+	# my $uid_min = UID_BLOCK * int($max/UID_BLOCK) + 1;
+}
+
+sub imapd_refresh_finalize {
+	my ($imapd, $pi_config) = @_;
+	$imapd->{groups} = delete $imapd->{tmp_groups};
+	$imapd->{grouplist} = [ values %{$imapd->{groups}} ];
+	refresh_inboxlist($imapd);
+	$imapd->{pi_config} = $pi_config;
+	if (my $idler = $imapd->{idler}) {
 		$idler->refresh($pi_config);
 	}
 }
 
+sub imapd_refresh_step { # pi_config->iterate_start cb
+	my ($pi_config, $section, $imapd) = @_;
+	if (defined($section)) {
+		return if $section !~ m!\Apublicinbox\.([^/]+)\z!;
+		my $ibx = $pi_config->lookup_name($1) or return;
+		imapd_refresh_ibx($ibx, $imapd);
+	} else { # "EOF"
+		imapd_refresh_finalize($imapd, $pi_config);
+	}
+}
+
+sub refresh_groups {
+	my ($self, $sig) = @_;
+	my $pi_config = PublicInbox::Config->new;
+	$self->{tmp_groups} = {};
+	if (0 && $sig) { # SIGHUP
+		$pi_config->iterate_start(\&imapd_refresh_step, $self);
+		PublicInbox::DS::requeue($pi_config); # call event_step
+	} else { # initial start
+		$pi_config->each_inbox(\&imapd_refresh_ibx, $self);
+		imapd_refresh_finalize($self, $pi_config);
+	}
+}
+
 sub idler_start {
 	$_[0]->{idler} //= PublicInbox::InboxIdle->new($_[0]->{pi_config});
 }
diff --git a/script/public-inbox-imapd b/script/public-inbox-imapd
index 63f865f53f0..60f2e6d8a36 100644
--- a/script/public-inbox-imapd
+++ b/script/public-inbox-imapd
@@ -9,6 +9,6 @@ use PublicInbox::IMAPdeflate; # loads PublicInbox::IMAP
 use PublicInbox::IMAPD;
 my $imapd = PublicInbox::IMAPD->new;
 PublicInbox::Daemon::run('0.0.0.0:143',
-	sub { $imapd->refresh_groups }, # refresh
+	sub { $imapd->refresh_groups(@_) }, # refresh
 	sub ($$$) { PublicInbox::IMAP->new($_[0], $imapd) }, # post_accept
 	$imapd);

  parent reply	other threads:[~2020-06-10  7:06 UTC|newest]

Thread overview: 84+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-06-10  7:03 [PATCH 00/82] public-inbox-imapd: read-only IMAP server Eric Wong
2020-06-10  7:03 ` [PATCH 01/82] doc: add some IMAP standards Eric Wong
2020-06-10  7:03 ` [PATCH 02/82] nntpd: restrict allowed newsgroup names Eric Wong
2020-06-10  7:04 ` [PATCH 03/82] preliminary imap server implementation Eric Wong
2020-06-10  7:04 ` [PATCH 04/82] inboxidle: new class to detect inbox changes Eric Wong
2020-06-10  7:04 ` [PATCH 05/82] imap: support IDLE Eric Wong
2020-06-10  7:04 ` [PATCH 06/82] msgmap: split ->max into its own method Eric Wong
2020-06-10  7:04 ` [PATCH 07/82] imap: delay InboxIdle start, support refresh Eric Wong
2020-06-10  7:04 ` [PATCH 08/82] imap: implement STATUS command Eric Wong
2020-06-10  7:04 ` [PATCH 09/82] imap: use Text::ParseWords::parse_line to handle quoted words Eric Wong
2020-06-10  7:04 ` [PATCH 10/82] imap: support LIST command Eric Wong
2020-06-10  7:04 ` [PATCH 11/82] t/imapd: support FakeInotify and KQNotify Eric Wong
2020-06-10  7:04 ` [PATCH 12/82] imap: support fetch for BODYSTRUCTURE and BODY Eric Wong
2020-06-10  7:04 ` [PATCH 13/82] eml: each_part: single part $idx is 1 Eric Wong
2020-06-10  7:04 ` [PATCH 14/82] imap: allow fetch of partial of BODY[...] and headers Eric Wong
2020-06-10  7:04 ` [PATCH 15/82] imap: always include `resp-text' in responses Eric Wong
2020-06-10  7:04 ` [PATCH 16/82] imap: split out unit tests and benchmarks Eric Wong
2020-06-10  7:04 ` [PATCH 17/82] imap: fix multi-message partial header fetches Eric Wong
2020-06-10  7:04 ` [PATCH 18/82] imap: simplify partial fetch structure Eric Wong
2020-06-10  7:04 ` [PATCH 19/82] imap: support sequence number FETCH Eric Wong
2020-06-10  7:04 ` [PATCH 20/82] imap: do not include ".PEEK" in responses Eric Wong
2020-06-10  7:04 ` [PATCH 21/82] imap: support the CLOSE command Eric Wong
2020-06-10  7:04 ` [PATCH 22/82] imap: speed up HEADER.FIELDS[.NOT] range fetches Eric Wong
2020-06-10  7:04 ` [PATCH 23/82] git: async: flatten the inflight array Eric Wong
2020-06-10  7:04 ` [PATCH 24/82] git: do our own read buffering for cat-file Eric Wong
2020-06-10  7:04 ` [PATCH 25/82] imap: use git-cat-file asynchronously Eric Wong
2020-06-10  7:04 ` [PATCH 26/82] git: idle rbuf for async Eric Wong
2020-06-10  7:04 ` [PATCH 27/82] imap: support LSUB command Eric Wong
2020-06-10  7:04 ` [PATCH 28/82] imap: FETCH: support comma-delimited ranges Eric Wong
2020-06-10  7:04 ` [PATCH 29/82] add imapd compression test Eric Wong
2020-06-10  7:04 ` [PATCH 30/82] testcommon: tcp_(server|connect): BAIL_OUT on failure Eric Wong
2020-06-10  7:04 ` [PATCH 31/82] *deflate: drop invalid comment about rbuf Eric Wong
2020-06-10  7:04 ` [PATCH 32/82] imap: fix pipelining with async git Eric Wong
2020-06-10  7:04 ` [PATCH 33/82] git: cat_async: provide requested OID + "missing" on missing blobs Eric Wong
2020-06-10  7:04 ` [PATCH 34/82] git: move async_cat reference to PublicInbox::Git Eric Wong
2020-06-10  7:04 ` [PATCH 35/82] git: async: automatic retry on alternates change Eric Wong
2020-06-10  7:04 ` [PATCH 36/82] imapclient: wrapper for Mail::IMAPClient Eric Wong
2020-06-10  7:04 ` [PATCH 37/82] xt: add imapd-validate and imapd-mbsync-oimap Eric Wong
2020-06-10  7:04 ` [PATCH 38/82] imap: support out-of-bounds ranges Eric Wong
2020-06-10  7:04 ` [PATCH 39/82] xt/perf-imap-list: time refresh_inboxlist Eric Wong
2020-06-10  7:04 ` [PATCH 40/82] imap: case-insensitive mailbox name comparisons Eric Wong
2020-06-10  7:04 ` [PATCH 41/82] imap: break giant inboxes into sub-inboxes of 50K messages Eric Wong
2020-06-10  7:04 ` Eric Wong [this message]
2020-06-10  7:04 ` [PATCH 43/82] imap: require ".$UID_MIN-$UID_END" suffix Eric Wong
2020-06-10  7:04 ` [PATCH 44/82] imapd: ensure LIST is sorted alphabetically, for now Eric Wong
2020-06-10  7:04 ` [PATCH 45/82] imap: omit $UID_END from mailbox name, use index Eric Wong
2020-06-10  7:04 ` [PATCH 46/82] t/config.t: always compare against git bool behavior Eric Wong
2020-06-10  7:04 ` [PATCH 47/82] xt/*: show some tunable parameters Eric Wong
2020-06-10  7:04 ` [PATCH 48/82] imap: STATUS and LIST are case-insensitive, too Eric Wong
2020-06-10  7:04 ` [PATCH 49/82] imap: EXAMINE/STATUS: return correct counts Eric Wong
2020-06-10  7:04 ` [PATCH 50/82] imap: avoid uninitialized warnings on incomplete commands Eric Wong
2020-06-10  7:04 ` [PATCH 51/82] imap: start parsing out queries for SQLite and Xapian Eric Wong
2020-06-10  7:04 ` [PATCH 52/82] imap: SEARCH: clamp results to the 50K UID range Eric Wong
2020-06-10  7:04 ` [PATCH 53/82] imap: allow UID range search on timestamps Eric Wong
2020-06-10  7:04 ` [PATCH 54/82] over: get_art: use dbh->prepare_cached Eric Wong
2020-06-10  7:04 ` [PATCH 55/82] search: index byte size of a message for IMAP search Eric Wong
2020-06-10  7:04 ` [PATCH 56/82] search: index UID for IMAP search, too Eric Wong
2020-06-10  7:04 ` [PATCH 57/82] imap: remove dummies from sequence number FETCH Eric Wong
2020-06-10  7:04 ` [PATCH 58/82] imap: compile UID FETCH to opcodes Eric Wong
2020-06-10  7:04 ` [PATCH 59/82] imap: UID FETCH: optimize for smsg-only case Eric Wong
2020-06-10  7:04 ` [PATCH 60/82] imap: UID FETCH: optimize (UID FLAGS) harder Eric Wong
2020-06-10  7:04 ` [PATCH 61/82] imap: IDLE: avoid extraneous wakeups, keep-alive Eric Wong
2020-06-10  7:04 ` [PATCH 62/82] imap: 30 minute auto-logout timer Eric Wong
2020-06-10  7:05 ` [PATCH 63/82] imap: split ->logged_in attribute into a separate class Eric Wong
2020-06-10  7:05 ` [PATCH 64/82] searchidx: v1 (re)-index uses git asynchronously Eric Wong
2020-06-10  7:05 ` [PATCH 65/82] index: account for CRLF conversion when storing bytes Eric Wong
2020-06-10  7:05 ` [PATCH 66/82] imap: rely on smsg->{bytes} for RFC822.SIZE Eric Wong
2020-06-10  7:05 ` [PATCH 67/82] imap: UID FETCH requires at least one data item Eric Wong
2020-06-10  7:05 ` [PATCH 68/82] imap: LIST shows "INBOX" in all caps Eric Wong
2020-06-10  7:05 ` [PATCH 69/82] imap: support 8000 octet lines Eric Wong
2020-06-10  7:05 ` [PATCH 70/82] imap: reinstate some message sequence number support Eric Wong
2020-06-10  7:05 ` [PATCH 71/82] imap: cleanup ->{uid_base} usage Eric Wong
2020-06-10  7:05 ` [PATCH 72/82] imap: FETCH: more granular CRLF conversion Eric Wong
2020-06-10  7:05 ` [PATCH 73/82] imap: further speed up HEADER.FIELDS FETCH requests Eric Wong
2020-06-10  7:05 ` [PATCH 74/82] imap: FETCH: try to make fake MSNs sequentially Eric Wong
2020-06-10  7:05 ` [PATCH 75/82] imap: STATUS/EXAMINE: rely on SQLite overview Eric Wong
2020-06-10  7:05 ` [PATCH 76/82] imap: UID SEARCH: support multiple ranges Eric Wong
2020-06-10  7:05 ` [PATCH 77/82] imap: wire up Xapian, MSN SEARCH and multi sequence-sets Eric Wong
2020-06-10  7:05 ` [PATCH 78/82] imap: misc cleanups and notes Eric Wong
2020-06-10  7:05 ` [PATCH 79/82] imapd: don't bother sorting LIST output Eric Wong
2020-06-10  7:05 ` [PATCH 80/82] imap: remove non-UID SEARCH for now Eric Wong
2020-06-10  7:05 ` [PATCH 81/82] over: uid_range: remove LIMIT Eric Wong
2020-06-10  7:05 ` [PATCH 82/82] imap: FETCH: proper MSN => UID mapping for requests Eric Wong
2020-06-12 23:49 ` [PATCH 83/82] imap: introduce memory-efficient uo2m mapping Eric Wong

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://public-inbox.org/README

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200610070519.18252-43-e@yhbt.net \
    --to=e@yhbt.net \
    --cc=meta@public-inbox.org \
    --subject='Re: [PATCH 42/82] imap: start doing iterative config reloading' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

Code repositories for project(s) associated with this inbox:

	https://80x24.org/public-inbox.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).