From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <e@80x24.org>
X-Spam-Checker-Version: SpamAssassin 3.4.2 (2018-09-13) on dcvr.yhbt.net
X-Spam-Level: 
X-Spam-ASN:  
X-Spam-Status: No, score=-4.0 required=3.0 tests=ALL_TRUSTED,AWL,BAYES_00,
	DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,DKIM_VALID_EF shortcircuit=no
	autolearn=ham autolearn_force=no version=3.4.2
Received: from localhost (dcvr.yhbt.net [127.0.0.1])
	by dcvr.yhbt.net (Postfix) with ESMTP id 0BCAC1F51E
	for <meta@public-inbox.org>; Fri, 30 Sep 2022 17:20:35 +0000 (UTC)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=80x24.org;
	s=selector1; t=1664558435;
	bh=ViAm+13I2uC2efCwhmCG06iybFQzrRpOYcbDhCfFGWs=;
	h=Date:From:To:Subject:References:In-Reply-To:From;
	b=awmNhABmr9a5A0uq6akVOYLH8/wNLVc1wnTsUfwjMh59ekJgLMWPVV4B9M326nERG
	 mlroHnfzRpB2DZEEk4q/++8bTCeZRTJmGxheogH3UWHxjh37qR+HYJORlIIewBT1Dw
	 5m6sH+628cQu3ZR+hJ1nW0gzqp+h+dvED6ESWqeM=
Date: Fri, 30 Sep 2022 17:20:34 +0000
From: Eric Wong <e@80x24.org>
To: meta@public-inbox.org
Subject: SQLite <3.8.3 was broken on fork (was: fixes noticed while
 diagnosing t/lei-up.t)
Message-ID: <20220930172034.GA24156@dcvr>
References: <20220930092140.3894047-1-e@80x24.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20220930092140.3894047-1-e@80x24.org>
List-Id: <meta.public-inbox.org>

Eric Wong <e@80x24.org> wrote:
> I'm still trying to figure out why OverIdx->adj_counter (via
> next_tid) in the LeiSavedSearch dedupe check occasionally fails
> under CentOS 7.x (but not other systems).

CentOS 7.x ships SQLite 3.7.17, and SQLite prior to 3.8.3 (2014-02-03)
didn't reset its PRNG for generating temporary filenames upon fork.

I can't find a way reset that PRNG via DBD::SQLite, either...

One possible nasty thing I could do is reset TMPDIR upon fork,
but that may leave bugs lingering, too :<

Here's the debugging patch I used to strace what was going on:
(I may just make lei_ok bail out on error...)

diff --git a/lib/PublicInbox/OverIdx.pm b/lib/PublicInbox/OverIdx.pm
index e7c96e14..e1af910c 100644
--- a/lib/PublicInbox/OverIdx.pm
+++ b/lib/PublicInbox/OverIdx.pm
@@ -51,10 +51,22 @@ SELECT val FROM counter WHERE key = ? LIMIT 1
 sub adj_counter ($$$) {
 	my ($self, $key, $op) = @_;
 	my $dbh = $self->{dbh};
+use PublicInbox::Spawn qw(spawn);
+my $trace = "/tmp/$$.strace";
+open my $err, '>>', $trace;
+my $pid = spawn([qw(strace -f -s4096 -v -p), $$], undef, { 2 => $err });
+select undef, undef, undef, 0.1; 
+
 	my $sth = $dbh->prepare_cached(<<"");
 UPDATE counter SET val = val $op 1 WHERE key = ?
 
-	$sth->execute($key);
+	eval { $sth->execute($key) };
+my $err = $@;
+syswrite(STDERR, "$$ $err\n") if $err;
+kill('TERM', $pid);
+waitpid($pid, 0);
+die $err if $err;
+#unlink($trace);
 
 	get_counter($dbh, $key);
 }
diff --git a/t/lei-up.t b/t/lei-up.t
index baed6507..9c65a243 100644
--- a/t/lei-up.t
+++ b/t/lei-up.t
@@ -27,7 +27,7 @@ test_lei(sub {
 	lei_ok qw(ls-search);
 	$s = eml_load('t/utf8.eml')->as_string;
 	lei_ok [qw(import -q -F eml -)], undef, { 0 => \$s, %$lei_opt };
-	lei_ok qw(up --all=local);
+	lei_ok qw(up --all=local) or xbail "lei up --all=local failed $?";
 
 	gunzip("$home/a.mbox.gz" => \$uc, MultiStream => 1) or
 		 xbail "gunzip $GunzipError";