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-Status: No, score=-4.0 required=3.0 tests=ALL_TRUSTED,BAYES_00
	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 481151F66F;
	Tue, 10 Nov 2020 18:53:51 +0000 (UTC)
Date: Tue, 10 Nov 2020 18:53:51 +0000
From: Eric Wong <e@80x24.org>
To: meta@public-inbox.org
Subject: detached external index: performance note
Message-ID: <20201110185351.GA11848@dcvr>
References: <20201027075453.19163-1-e@80x24.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20201027075453.19163-1-e@80x24.org>
List-Id: <meta.public-inbox.org>

Eric Wong <e@80x24.org> wrote:
> Not sure about the usability aspects, but I think this can
> replace the need for per-inbox Xapian DBs and save a truckload
> of disk space (and more importantly: cache space).  Per-inbox
> over.sqlite3 remains required for compatibility with NNTP/IMAP
> and existing WWW code.

Keeping v2 indexlevel=basic (*.sqlite3) and git repos on HDD
and putting -extindex on SSD seems to work reasonably well.
Xapian on HDD is really painful.

> Performance isn't great, it took 30+ hours to index my mirror of
> lore on a SATA SSD, but the entire index is <200GB due to
> deduplication between cross posts.

Still a problem for RAM-starved systems and Xapian :<

Larger systems can use --batch-size, and maybe Sys::Meminfo
can be used (if installed) to determine a larger batch-size
by default.

Fortunately, Sys::Meminfo is packaged for FreeBSD, CentOS and
Debian so it can be an optional dependency.