user/dev discussion of public-inbox itself
 help / color / Atom feed
From: ebiederm@xmission.com (Eric W. Biederman)
To: Eric Wong <e@80x24.org>
Cc: meta@public-inbox.org
Subject: Re: [WIP] searchidx: support incremental indexing on indexlevel=basic
Date: Thu, 02 Aug 2018 13:15:49 -0500
Message-ID: <87bmakiqi2.fsf@xmission.com> (raw)
In-Reply-To: <87h8kcitfm.fsf@xmission.com>

ebiederm@xmission.com (Eric W. Biederman) writes:

> ebiederm@xmission.com (Eric W. Biederman) writes:
>
>> Eric Wong <e@80x24.org> writes:
>>
>>> I wrote:
>>>> While testing this, it looks like I introduced a bug to
>>>> indexlevel=basic which broke incremental indexing when I made it
>>>> possible to upgrade to (medium|full).  Patch coming for that in
>>>> a bit...
>>>
>>> Eep, I think there's deeper problems with indexlevel=basic and
>>> incremental updates, since it's still doing lookups against
>>> Xapian for deletes...
>>>
>>> This is my work-in-progress to stop full-reindexing, at least.
>>
>> So I don't think your change is quite right.
>>
>> If the repo has only lived with indexlevel == 'basic' then last_commit
>> should always match the output of xdb->get_metadata('last_commit') as
>> it will always been undef or ''.
>>
>> In fact I don't see any code that is updating
>> $xdb->get_metadata('last_commit');
>>
>> Am I wrong or can we just always use mm->last_commit for this logic?
>
> Yes.  I was wrong.  I had overlooked the get_metadata($mkey) and
> set_metadata($mkey) calls.   From just below but those are all
> with $mkey == 'last_commit'  So 'last_commit' does get updated.
>
> I think we want something more like below.  Where _last_x_commit
> just computes the last_commit value to use, using both values of
> last_commit.
>
> I am not 100% about the loop.  Perhaps that check to see if either
> value of last_commit changes?
>
> I definitely don't think when we already have one set of logic
> for substituting in last_commit we should be adding in another.
> That gets confusing very fast.
>
Make that something like this where we remove the $mkey variable 
and all of it's special semantics with regards to reindexing entirely.

Eric


diff --git a/lib/PublicInbox/SearchIdx.pm b/lib/PublicInbox/SearchIdx.pm
index 198f39b96e6c..8b12da0436a4 100644
--- a/lib/PublicInbox/SearchIdx.pm
+++ b/lib/PublicInbox/SearchIdx.pm
@@ -669,6 +669,23 @@ sub need_update ($$$) {
 	($n eq '' || $n > 0);
 }
 
+sub _last_x_commit
+{
+	my ($self) = @_;
+        my $lm = $mm->last_commit || '';
+	my $lx = '';
+	if ($self->{indexlevel} ~= $xapianlevels) {
+		$lx = $self->{xdb}->get_metadata('last_commit') || '';
+        } else {
+		$lx = $lm;
+	}
+	# Use last_commit from msgmap if it is older or unset
+	if (!$lm || ($lx && $lx && is_ancestor($git, $lm, $lx))) {
+		$lx = $lm;
+	}
+	$lx;
+}
+
 # indexes all unindexed messages (v1 only)
 sub _index_sync {
 	my ($self, $opts) = @_;
@@ -682,19 +699,8 @@ sub _index_sync {
 	my $mm = _msgmap_init($self);
 	do {
 		$xlog = undef;
-		$mkey = 'last_commit';
-		$last_commit = $xdb->get_metadata('last_commit');
-		$lx = $last_commit;
-		if ($reindex) {
-			$lx = '';
-			$mkey = undef if $last_commit ne '';
-		}
-
-		# use last_commit from msgmap if it is older or unset
-		my $lm = $mm->last_commit || '';
-		if (!$lm || ($lm && $lx && is_ancestor($git, $lm, $lx))) {
-			$lx = $lm;
-		}
+		$last_commit = $self->_last_x_commit;
+		$lx = !$reindex ? $last_commit : '';
 
 		$self->{over}->rollback_lazy;
 		$self->{over}->disconnect;
@@ -708,7 +714,7 @@ sub _index_sync {
 		$xlog = _git_log($self, $range);
 
 		$xdb = $self->begin_txn_lazy;
-	} while ($xdb->get_metadata('last_commit') ne $last_commit);
+	} while ($self->_last_x_commit ne $last_commit);
 
 	my $dbh = $mm->{dbh} if $mm;
 	my $cb = sub {
@@ -722,10 +728,10 @@ sub _index_sync {
 			}
 			$dbh->commit;
 		}
-		if ($mkey && $newest && $self->{indexlevel} =~ $xapianlevels) {
-			my $cur = $xdb->get_metadata($mkey);
+		if ($newest && $self->{indexlevel} =~ $xapianlevels) {
+			my $cur = $xdb->get_metadata('last_commit');
 			if (need_update($self, $cur, $newest)) {
-				$xdb->set_metadata($mkey, $newest);
+				$xdb->set_metadata('last_commit', $newest);
 			}
 		}
 		$self->commit_txn_lazy;

  reply index

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-08-01 16:41 [PATCH 00/13]: Incremental index fixes ebiederm
2018-08-01 16:43 ` [PATCH 01/13] Import.pm: Don't assume {in} and {out} always exist Eric W. Biederman
2018-08-01 16:43 ` [PATCH 02/13] t/v1reindex.t: Isolate the test cases Eric W. Biederman
2018-08-01 16:43 ` [PATCH 03/13] t/v2reindex.t: Isolate the test cases more Eric W. Biederman
2018-08-01 16:43 ` [PATCH 04/13] t/v[12]reindex.t: Place expected second in Xapian tests Eric W. Biederman
2018-08-01 16:43 ` [PATCH 05/13] t/v[12]reindex.t: Test that the resulting msgmap is as expected Eric W. Biederman
2018-08-01 16:43 ` [PATCH 06/13] t/v[12]reindex.t: Test incremental indexing works Eric W. Biederman
2018-08-01 16:43 ` [PATCH 07/13] SearchIdx.pm: Always assign numbers backwards during incremental indexing Eric W. Biederman
2018-08-01 16:43 ` [PATCH 08/13] Msgmap.pm: Track the largest value of num ever assigned Eric W. Biederman
2018-08-02  3:00   ` Eric Wong
2018-08-02  3:44     ` [WIP] searchidx: support incremental indexing on indexlevel=basic Eric Wong
2018-08-02 12:25       ` ebiederm
2018-08-02 17:12         ` ebiederm
2018-08-02 18:15           ` ebiederm [this message]
2019-05-14  2:04       ` [PATCH 0/3] incremental index fixes for indexlevel=basic Eric Wong
2019-05-14  2:04         ` [PATCH 1/3] v1writable: new wrapper which is closer to v2writable Eric Wong
2019-05-14  2:04         ` [PATCH 2/3] v2writable: allow setting nproc via creat options Eric Wong
2019-05-14  2:04         ` [PATCH 3/3] searchidx: fix incremental index with indexlevel=basic on v1 Eric Wong
2018-08-02 12:08     ` [PATCH 08/13] Msgmap.pm: Track the largest value of num ever assigned ebiederm
2018-08-01 16:43 ` [PATCH 09/13] t/v[12]reindex.t Verify num_highwater Eric W. Biederman
2018-08-01 16:43 ` [PATCH 10/13] t/v[12]reindex.t: Verify the num highwater is as expected Eric W. Biederman
2018-08-01 16:43 ` [PATCH 11/13] SearchIdx,V2Writeable: Update num_highwater on optimized deletes Eric W. Biederman
2018-08-01 16:43 ` [PATCH 12/13] V2Writeable.pm: Ensure that a found message number is in the msgmap Eric W. Biederman
2018-08-01 16:43 ` [PATCH 13/13] V2Writeable.pm: In unindex_oid delete the message from msgmap Eric W. Biederman

Reply instructions:

You may reply publically 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=87bmakiqi2.fsf@xmission.com \
    --to=ebiederm@xmission.com \
    --cc=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

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

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.org/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