We need to send the socket over to lei/store and wait for the
kernel to drop the socket refcount down to zero before
script/lei can exit.
This is not a new bug and only caused very sporadic test
failures. I only noticed it while simplifying IPC stuff.
---
lib/PublicInbox/LeiNoteEvent.pm | 15 ++++++++-------
1 file changed, 8 insertions(+), 7 deletions(-)
diff --git a/lib/PublicInbox/LeiNoteEvent.pm b/lib/PublicInbox/LeiNoteEvent.pm
index ba4dfd49..1749c98f 100644
--- a/lib/PublicInbox/LeiNoteEvent.pm
+++ b/lib/PublicInbox/LeiNoteEvent.pm
@@ -11,10 +11,11 @@ use PublicInbox::DS;
our $to_flush; # { cfgpath => $lei }
-sub flush_lei ($) {
- my ($lei) = @_;
- my $lne = delete $lei->{cfg}->{-lei_note_event};
- $lne->wq_close if $lne; # runs _lei_wq_eof;
+sub flush_lei ($;$) {
+ my ($lei, $manual) = @_;
+ my $lne = delete $lei->{cfg}->{-lei_note_event} // return;
+ $lne->{lei_sock} = $lei->{sock} if $manual;
+ $lne->wq_close; # runs _lei_wq_eof;
}
# we batch up writes and flush every 5s (matching Linux default
@@ -67,7 +68,7 @@ sub lei_note_event {
die "BUG: unexpected: @rest" if @rest;
my $cfg = $lei->_lei_cfg or return; # gone (race)
my $sto = $lei->_lei_store or return; # gone
- return flush_lei($lei) if $folder eq 'done'; # special case
+ return flush_lei($lei, 1) if $folder eq 'done'; # special case
my $lms = $lei->lms or return;
$lms->lms_write_prepare if $new_cur eq ''; # for ->clear_src below
$lei->{opt}->{quiet} = 1;
@@ -111,8 +112,8 @@ sub ipc_atfork_child {
sub _lei_wq_eof { # EOF callback for main lei daemon
my ($lei) = @_;
- delete $lei->{lne} or return $lei->fail;
- $lei->sto_done_request;
+ my $lne = delete $lei->{lne} or return $lei->fail;
+ $lei->sto_done_request($lne->{lei_sock});
}
1;
--
unsubscribe: one-click, see List-Unsubscribe header
archive: https://public-inbox.org/meta/
