user/dev discussion of public-inbox itself
 help / color / mirror / code / Atom feed
From: Eric Wong <e@80x24.org>
To: meta@public-inbox.org
Subject: [PATCH 10/17] daemon: avoid race when quitting workers
Date: Sun, 24 Nov 2019 00:22:30 +0000	[thread overview]
Message-ID: <20191124002237.15713-11-e@80x24.org> (raw)
In-Reply-To: <20191124002237.15713-1-e@80x24.org>

While the master process has a self-pipe to avoid missing
signals, worker processes lack that aside from a pipe to
detect master death.

That pipe doesn't exist when there's no master process,
so it's possible DS::close never finishes because it
never woke up from epoll_wait.  So create a pipe on
the worker_quit signal and force it into epoll/kevent
so it wakes up right away.
---
 lib/PublicInbox/Daemon.pm | 21 +++++++++++++++++----
 1 file changed, 17 insertions(+), 4 deletions(-)

diff --git a/lib/PublicInbox/Daemon.pm b/lib/PublicInbox/Daemon.pm
index 90f11137..0e3b95d2 100644
--- a/lib/PublicInbox/Daemon.pm
+++ b/lib/PublicInbox/Daemon.pm
@@ -252,6 +252,11 @@ sub daemonize () {
 	}
 }
 
+sub shrink_pipes {
+	if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ, 4096: page size
+		fcntl($_, 1031, 4096) for @_;
+	}
+}
 
 sub worker_quit {
 	# killing again terminates immediately:
@@ -260,6 +265,17 @@ sub worker_quit {
 	$_->close foreach @listeners; # call PublicInbox::DS::close
 	@listeners = ();
 
+	# create a lazy self-pipe which kicks us out of the EventLoop
+	# so DS::PostEventLoop can fire
+	if (pipe(my ($r, $w))) {
+		shrink_pipes($w);
+
+		# shrink_pipes == noop
+		PublicInbox::ParentPipe->new($r, *shrink_pipes);
+		close $w; # wake up from the event loop
+	} else {
+		warn "E: pipe failed ($!), quit unreliable\n";
+	}
 	my $proc_name;
 	my $warn = 0;
 	# drop idle connections and try to quit gracefully
@@ -468,10 +484,7 @@ sub unlink_pid_file_safe_ish ($$) {
 sub master_loop {
 	pipe(my ($p0, $p1)) or die "failed to create parent-pipe: $!";
 	pipe(my ($r, $w)) or die "failed to create self-pipe: $!";
-
-	if ($^O eq 'linux') { # 1031: F_SETPIPE_SZ = 1031
-		fcntl($_, 1031, 4096) for ($w, $p1);
-	}
+	shrink_pipes($w, $p1);
 
 	IO::Handle::blocking($w, 0);
 	my $set_workers = $worker_processes;

  parent reply	other threads:[~2019-11-24  0:22 UTC|newest]

Thread overview: 54+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-15  9:50 [PATCH 00/29] speed up tests by preloading Eric Wong
2019-11-15  9:50 ` [PATCH 01/29] edit: pass global variables into subs Eric Wong
2019-11-15  9:50 ` [PATCH 02/29] edit: use OO API of File::Temp to shorten lifetime Eric Wong
2019-11-15  9:50 ` [PATCH 03/29] admin: get rid of singleton $CFG var Eric Wong
2019-11-15  9:50 ` [PATCH 04/29] index: pass global variables into subs Eric Wong
2019-11-15  9:50 ` [PATCH 05/29] init: " Eric Wong
2019-11-15  9:50 ` [PATCH 06/29] mda: " Eric Wong
2019-11-15  9:50 ` [PATCH 07/29] learn: " Eric Wong
2019-11-15  9:50 ` [PATCH 08/29] inboxwritable: add ->cleanup method Eric Wong
2019-11-15  9:50 ` [PATCH 09/29] import: only pass Inbox object to SearchIdx->new Eric Wong
2019-11-15  9:50 ` [PATCH 10/29] xapcmd: do not fire END and DESTROY handlers in child Eric Wong
2019-11-15  9:50 ` [PATCH 11/29] spawn: which: allow embedded slash for relative path Eric Wong
2019-11-15  9:50 ` [PATCH 12/29] t/common: introduce run_script wrapper for t/cgi.t Eric Wong
2019-11-15  9:50 ` [PATCH 13/29] t/edit: switch to use run_script Eric Wong
2019-11-15  9:50 ` [PATCH 14/29] t/init: convert to using run_script Eric Wong
2019-11-15  9:50 ` [PATCH 15/29] t/purge: convert to run_script Eric Wong
2019-11-15  9:50 ` [PATCH 16/29] t/v2mirror: get rid of IPC::Run dependency Eric Wong
2019-11-15  9:50 ` [PATCH 17/29] t/mda: switch to run_script for testing Eric Wong
2019-11-15  9:50 ` [PATCH 18/29] t/mda_filter_rubylang: drop IPC::Run dependency Eric Wong
2019-11-15  9:50 ` [PATCH 19/29] doc: remove IPC::Run as a dev and test dependency Eric Wong
2019-11-15  9:50 ` [PATCH 20/29] t/v2mirror: switch to default run_mode for speedup Eric Wong
2019-11-15  9:50 ` [PATCH 21/29] t/convert-compact: convert to run_script Eric Wong
2019-11-15  9:50 ` [PATCH 22/29] t/httpd: use run_script for -init Eric Wong
2019-11-15  9:50 ` [PATCH 23/29] t/watch_maildir_v2: " Eric Wong
2019-11-15  9:50 ` [PATCH 24/29] t/nntpd: " Eric Wong
2019-11-15  9:50 ` [PATCH 25/29] t/watch_filter_rubylang: run_script for -init and -index Eric Wong
2019-11-15  9:50 ` [PATCH 26/29] t/v2mda: switch to run_script in many places Eric Wong
2019-11-15  9:50 ` [PATCH 27/29] t/indexlevels-mirror*: switch to run_script Eric Wong
2019-11-15  9:50 ` [PATCH 28/29] t/xcpdb-reshard: use run_script for -xcpdb Eric Wong
2019-11-15  9:51 ` [PATCH 29/29] t/common: start_script replaces spawn_listener Eric Wong
2019-11-16  6:52   ` Eric Wong
2019-11-16 11:43     ` Eric Wong
2019-11-24  0:22       ` [PATCH 00/17] test fixes and cleanups Eric Wong
2019-11-24  0:22         ` [PATCH 01/17] tests: disable daemon workers in a few more places Eric Wong
2019-11-24  0:22         ` [PATCH 02/17] tests: use strict everywhere Eric Wong
2019-11-24  0:22         ` [PATCH 03/17] t/v1-add-remove-add: quiet down "git init" Eric Wong
2019-11-24  0:22         ` [PATCH 04/17] t/xcpdb-reshard: test xcpdb --compact Eric Wong
2019-11-24  0:22         ` [PATCH 05/17] t/httpd-corner: wait for worker process death Eric Wong
2019-11-24  0:22         ` [PATCH 06/17] t/nntpd-tls: sometimes SSL_connect succeeds quickly Eric Wong
2019-11-24  0:22         ` [PATCH 07/17] .gitignore: ignore local prove(1) files Eric Wong
2019-11-24  0:22         ` [PATCH 08/17] daemon: use sigprocmask to block signals at startup Eric Wong
2019-11-24  0:22         ` [PATCH 09/17] daemon: use sigprocmask when respawning workers Eric Wong
2019-11-24  0:22         ` Eric Wong [this message]
2019-11-25  8:59           ` [PATCH 10/17] daemon: avoid race when quitting workers Eric Wong
2019-11-27  1:33             ` [PATCH 0/2] fix kqueue support and missed signal wakeups Eric Wong
2019-11-27  1:33               ` [PATCH 1/2] dskqxs: fix missing EV_DISPATCH define Eric Wong
2019-11-27  1:33               ` [PATCH 2/2] httpd|nntpd: avoid missed signal wakeups Eric Wong
2019-11-24  0:22         ` [PATCH 11/17] t/common: start_script replaces spawn_listener Eric Wong
2019-11-24  0:22         ` [PATCH 12/17] t/nntpd-validate: get rid of threads dependency Eric Wong
2019-11-24  0:22         ` [PATCH 13/17] xapcmd: replace Xtmpdirs with File::Temp->newdir Eric Wong
2019-11-24  0:22         ` [PATCH 14/17] tests: use File::Temp->newdir instead of tempdir() Eric Wong
2019-11-24  0:22         ` [PATCH 15/17] tests: quiet down commit graph Eric Wong
2019-11-24  0:22         ` [PATCH 16/17] t/perf-*.t: use $ENV{GIANT_INBOX_DIR} consistently Eric Wong
2019-11-24  0:22         ` [PATCH 17/17] tests: move giant inbox/git dependent tests to xt/ 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=20191124002237.15713-11-e@80x24.org \
    --to=e@80x24.org \
    --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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public 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).