user/dev discussion of public-inbox itself
 help / color / Atom feed
From: Eric Wong <e@yhbt.net>
To: meta@public-inbox.org
Subject: [PATCH 33/34] imaptracker: use flock(2) around writes
Date: Sat, 27 Jun 2020 10:03:59 +0000
Message-ID: <20200627100400.9871-34-e@yhbt.net> (raw)
In-Reply-To: <20200627100400.9871-1-e@yhbt.net>

SQLite only issues non-blocking F_SETLK ops (not F_SETLKW) and
retries failures using a configurable busy_timeout.  SQLite's
busy loop sleeps for a millisecond and retries the lock until
the configured busy_timeout is hit.

Trying to set ->sqlite_busy_timeout to larger values (e.g. 30000
milliseconds) still leads to failure when running the new stress
test with 8 processes with TMPDIR on a 7200 RPM HDD.

Inspection of SQLite source reveals there's no built-in way to
use F_SETLKW, so tack on the existing flock(2) support we use to
synchronize git + SQLite + Xapian for inbox writing.  We use
flock(2) instead of POSIX fcntl(2) locks since Perl doesn't
provide a way to manipulate "struct flock" portably.
---
 lib/PublicInbox/IMAPTracker.pm | 13 ++++++++++---
 t/imap_tracker.t               | 30 +++++++++++++++++++++++++++++-
 2 files changed, 39 insertions(+), 4 deletions(-)

diff --git a/lib/PublicInbox/IMAPTracker.pm b/lib/PublicInbox/IMAPTracker.pm
index 0bbabe07fae..102a74ce66b 100644
--- a/lib/PublicInbox/IMAPTracker.pm
+++ b/lib/PublicInbox/IMAPTracker.pm
@@ -2,6 +2,7 @@
 # License: AGPL-3.0+ <https://www.gnu.org/licenses/agpl-3.0.txt>
 package PublicInbox::IMAPTracker;
 use strict;
+use parent qw(PublicInbox::Lock);
 use DBI;
 use DBD::SQLite;
 use PublicInbox::Config;
@@ -48,7 +49,10 @@ sub update_last ($$$) {
 INSERT OR REPLACE INTO imap_last (url, uid_validity, uid)
 VALUES (?, ?, ?)
 
-	$sth->execute($self->{url}, $validity, $last);
+	$self->lock_acquire;
+	my $rv = $sth->execute($self->{url}, $validity, $last);
+	$self->lock_release;
+	$rv;
 }
 
 sub new {
@@ -68,8 +72,11 @@ sub new {
 		require File::Basename;
 		File::Path::mkpath(File::Basename::dirname($dbname));
 	}
-
-	bless { url => $url, dbh => dbh_new($dbname) }, $class;
+	my $self = bless { lock_path => "$dbname.lock", url => $url }, $class;
+	$self->lock_acquire;
+	$self->{dbh} = dbh_new($dbname);
+	$self->lock_release;
+	$self;
 }
 
 1;
diff --git a/t/imap_tracker.t b/t/imap_tracker.t
index 8dc04ed77a3..01e1d0b1549 100644
--- a/t/imap_tracker.t
+++ b/t/imap_tracker.t
@@ -9,8 +9,8 @@ my ($tmpdir, $for_destroy) = tmpdir();
 mkdir "$tmpdir/old" or die "mkdir $tmpdir/old: $!";
 my $old = "$tmpdir/old/imap.sqlite3";
 my $cur = "$tmpdir/data/public-inbox/imap.sqlite3";
+local $ENV{XDG_DATA_HOME} = "$tmpdir/data";
 {
-	local $ENV{XDG_DATA_HOME} = "$tmpdir/data";
 	local $ENV{PI_DIR} = "$tmpdir/old";
 
 	my $tracker = PublicInbox::IMAPTracker->new;
@@ -22,5 +22,33 @@ my $cur = "$tmpdir/data/public-inbox/imap.sqlite3";
 	$tracker = PublicInbox::IMAPTracker->new;
 	ok(!-f $cur, '->new does not create new file if old is present');
 }
+SKIP: {
+	my $nproc = $ENV{TEST_STRESS_NPROC};
+	skip 'TEST_STRESS_NPROC= not set', 1 unless $nproc;
+	my $nr = $ENV{TEST_STRESS_NR} // 10000;
+	diag "TEST_STRESS_NPROC=$nproc TEST_STRESS_NR=$nr";
+	require POSIX;
+	for my $n (1..$nproc) {
+		defined(my $pid = fork) or BAIL_OUT "fork: $!";
+		if ($pid == 0) {
+			my $url = "imap://example.com/INBOX.$$";
+			my $uidval = time;
+			eval {
+				my $itrk = PublicInbox::IMAPTracker->new($url);
+				for my $uid (1..$nr) {
+					$itrk->update_last($uidval, $uid);
+					my ($uv, $u) = $itrk->get_last;
+				}
+			};
+			warn "E: $n $$ - $@\n" if $@;
+			POSIX::_exit($@ ? 1 : 0);
+		}
+	}
+	while (1) {
+		my $pid = waitpid(-1, 0);
+		last if $pid < 0;
+		is($?, 0, "$pid exited");
+	}
+}
 
 done_testing;

  parent reply index

Thread overview: 46+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-06-27 10:03 [PATCH 00/34] watch: add IMAP and NNTP support Eric Wong
2020-06-27 10:03 ` [PATCH 01/34] inboxwritable: ensure ssoma.lock exists on init Eric Wong
2020-06-27 10:03 ` [PATCH 02/34] inbox: warn on ->on_inbox_unlock exception Eric Wong
2020-06-27 10:03 ` [PATCH 03/34] IMAPTracker: Add a helper to track our place in reading imap mailboxes Eric Wong
2020-06-27 10:03 ` [PATCH 04/34] imaptracker: use ~/.local/share/public-inbox/imap.sqlite3 Eric Wong
2020-06-27 10:03 ` [PATCH 05/34] watchmaildir: hoist out compile_watchheaders Eric Wong
2020-06-27 10:03 ` [PATCH 06/34] watchmaildir: fix check for spam vs ham inbox conflicts Eric Wong
2020-06-27 10:03 ` [PATCH 07/34] URI IMAP support Eric Wong
2020-06-27 10:03 ` [PATCH 08/34] watch: preliminary " Eric Wong
2020-06-27 10:03 ` [PATCH 09/34] kqnotify|fake_inotify: detect Maildir write ops Eric Wong
2020-06-27 10:03 ` [PATCH 10/34] watch: remove Filesys::Notify::Simple dependency Eric Wong
2020-06-27 10:03 ` [PATCH 11/34] watch: use signalfd for Maildir watching Eric Wong
2020-06-27 19:05   ` Kyle Meyer
2020-06-27 22:32     ` Eric Wong
2020-06-27 10:03 ` [PATCH 12/34] ds: remove fields.pm usage Eric Wong
2020-06-27 10:03 ` [PATCH 13/34] watch: wire up IMAP IDLE reapers to DS Eric Wong
2020-06-27 10:03 ` [PATCH 14/34] watch: support IMAP polling Eric Wong
2020-06-27 10:03 ` [PATCH 15/34] config: support ->urlmatch method for -watch Eric Wong
2020-06-27 10:03 ` [PATCH 16/34] watch: stop importers before forking Eric Wong
2020-06-27 10:03 ` [PATCH 17/34] watch: use UID SEARCH to avoid empty UID FETCH Eric Wong
2020-06-27 10:03 ` [PATCH 18/34] ds: add_timer: allow passing arg to callback Eric Wong
2020-06-27 10:03 ` [PATCH 19/34] imaptracker: add {url} field to reduce args Eric Wong
2020-06-27 10:03 ` [PATCH 20/34] imaptracker: drop {dbname} field Eric Wong
2020-06-27 10:03 ` [PATCH 21/34] watch: avoid long transaction to IMAPTracker Eric Wong
2020-06-27 10:03 ` [PATCH 22/34] watch: support imap.fetchBatchSize parameter Eric Wong
2020-06-27 10:03 ` [PATCH 23/34] watch: imap: be quiet about disconnecting on quit Eric Wong
2020-06-27 10:03 ` [PATCH 24/34] watch: support multiple watch: directives per-inbox Eric Wong
2020-06-27 10:03 ` [PATCH 25/34] watch: remove {mdir} array Eric Wong
2020-06-27 10:03 ` [PATCH 26/34] watch: just use ->urlmatch Eric Wong
2020-06-27 10:03 ` [PATCH 27/34] testcommon: $ENV{TAIL} supports non-@ARGV redirects Eric Wong
2020-06-27 10:03 ` [PATCH 28/34] watch: add NNTP support Eric Wong
2020-06-27 19:06   ` Kyle Meyer
2020-06-27 10:03 ` [PATCH 29/34] watch: show user-specified URL consistently Eric Wong
2020-06-27 10:03 ` [PATCH 30/34] watch: enable autoflush for STDOUT and STDERR Eric Wong
2020-06-27 10:03 ` [PATCH 31/34] watch: use our own "git credential" wrapper Eric Wong
2020-06-27 10:03 ` [PATCH 32/34] watch: support ~/.netrc via Net::Netrc Eric Wong
2020-06-27 10:03 ` Eric Wong [this message]
2020-06-27 10:04 ` [PATCH 34/34] watch: simplify internal structures Eric Wong
2020-06-29 10:34 ` [PATCH 0/5] watch: Maildir fixes Eric Wong
2020-06-29 10:34   ` [PATCH 1/5] watch: check for duplicates in ->over before spamcheck Eric Wong
2020-06-29 10:34   ` [PATCH 2/5] watch: show path for warnings from spam messages Eric Wong
2020-06-29 10:34   ` [PATCH 3/5] watch: ensure SIGCHLD works in forked children Eric Wong
2020-06-29 10:34   ` [PATCH 4/5] spawn: unblock SIGCHLD in subprocess Eric Wong
2020-07-07  6:17     ` [PATCH 6/5] t/spawn: fix test reliability Eric Wong
2020-06-29 10:34   ` [PATCH 5/5] watch: make waitpid() synchronous for Maildir scans Eric Wong
2020-06-29 10:37     ` 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: http://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=20200627100400.9871-34-e@yhbt.net \
    --to=e@yhbt.net \
    --cc=meta@public-inbox.org \
    /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

user/dev discussion of public-inbox itself

Archives are clonable:
	git clone --mirror http://public-inbox.org/meta
	git clone --mirror http://czquwvybam4bgbro.onion/meta
	git clone --mirror http://hjrcffqmbrq6wope.onion/meta
	git clone --mirror http://ou63pmih66umazou.onion/meta

Example config snippet for mirrors

Newsgroups are available over NNTP:
	nntp://news.public-inbox.org/inbox.comp.mail.public-inbox.meta
	nntp://ou63pmih66umazou.onion/inbox.comp.mail.public-inbox.meta
	nntp://czquwvybam4bgbro.onion/inbox.comp.mail.public-inbox.meta
	nntp://hjrcffqmbrq6wope.onion/inbox.comp.mail.public-inbox.meta
	nntp://news.gmane.io/gmane.mail.public-inbox.general

 note: .onion URLs require Tor: https://www.torproject.org/

AGPL code for this site: git clone https://public-inbox.org/public-inbox.git