<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN"
"http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
<head>
<meta http-equiv="Content-Type" content="application/xhtml+xml; charset=UTF-8" />
<meta name="generator" content="AsciiDoc 8.6.9" />
<title>Converting Recoll indexing to multithreading</title>
<style type="text/css">
/* Shared CSS for AsciiDoc xhtml11 and html5 backends */
/* Default font. */
body {
font-family: Georgia,serif;
}
/* Title font. */
h1, h2, h3, h4, h5, h6,
div.title, caption.title,
thead, p.table.header,
#toctitle,
#author, #revnumber, #revdate, #revremark,
#footer {
font-family: Arial,Helvetica,sans-serif;
}
body {
margin: 1em 5% 1em 5%;
}
a {
color: blue;
text-decoration: underline;
}
a:visited {
color: fuchsia;
}
em {
font-style: italic;
color: navy;
}
strong {
font-weight: bold;
color: #083194;
}
h1, h2, h3, h4, h5, h6 {
color: #527bbd;
margin-top: 1.2em;
margin-bottom: 0.5em;
line-height: 1.3;
}
h1, h2, h3 {
border-bottom: 2px solid silver;
}
h2 {
padding-top: 0.5em;
}
h3 {
float: left;
}
h3 + * {
clear: left;
}
h5 {
font-size: 1.0em;
}
div.sectionbody {
margin-left: 0;
}
hr {
border: 1px solid silver;
}
p {
margin-top: 0.5em;
margin-bottom: 0.5em;
}
ul, ol, li > p {
margin-top: 0;
}
ul > li { color: #aaa; }
ul > li > * { color: black; }
.monospaced, code, pre {
font-family: "Courier New", Courier, monospace;
font-size: inherit;
color: navy;
padding: 0;
margin: 0;
}
pre {
white-space: pre-wrap;
}
#author {
color: #527bbd;
font-weight: bold;
font-size: 1.1em;
}
#email {
}
#revnumber, #revdate, #revremark {
}
#footer {
font-size: small;
border-top: 2px solid silver;
padding-top: 0.5em;
margin-top: 4.0em;
}
#footer-text {
float: left;
padding-bottom: 0.5em;
}
#footer-badges {
float: right;
padding-bottom: 0.5em;
}
#preamble {
margin-top: 1.5em;
margin-bottom: 1.5em;
}
div.imageblock, div.exampleblock, div.verseblock,
div.quoteblock, div.literalblock, div.listingblock, div.sidebarblock,
div.admonitionblock {
margin-top: 1.0em;
margin-bottom: 1.5em;
}
div.admonitionblock {
margin-top: 2.0em;
margin-bottom: 2.0em;
margin-right: 10%;
color: #606060;
}
div.content { /* Block element content. */
padding: 0;
}
/* Block element titles. */
div.title, caption.title {
color: #527bbd;
font-weight: bold;
text-align: left;
margin-top: 1.0em;
margin-bottom: 0.5em;
}
div.title + * {
margin-top: 0;
}
td div.title:first-child {
margin-top: 0.0em;
}
div.content div.title:first-child {
margin-top: 0.0em;
}
div.content + div.title {
margin-top: 0.0em;
}
div.sidebarblock > div.content {
background: #ffffee;
border: 1px solid #dddddd;
border-left: 4px solid #f0f0f0;
padding: 0.5em;
}
div.listingblock > div.content {
border: 1px solid #dddddd;
border-left: 5px solid #f0f0f0;
background: #f8f8f8;
padding: 0.5em;
}
div.quoteblock, div.verseblock {
padding-left: 1.0em;
margin-left: 1.0em;
margin-right: 10%;
border-left: 5px solid #f0f0f0;
color: #888;
}
div.quoteblock > div.attribution {
padding-top: 0.5em;
text-align: right;
}
div.verseblock > pre.content {
font-family: inherit;
font-size: inherit;
}
div.verseblock > div.attribution {
padding-top: 0.75em;
text-align: left;
}
/* DEPRECATED: Pre version 8.2.7 verse style literal block. */
div.verseblock + div.attribution {
text-align: left;
}
div.admonitionblock .icon {
vertical-align: top;
font-size: 1.1em;
font-weight: bold;
text-decoration: underline;
color: #527bbd;
padding-right: 0.5em;
}
div.admonitionblock td.content {
padding-left: 0.5em;
border-left: 3px solid #dddddd;
}
div.exampleblock > div.content {
border-left: 3px solid #dddddd;
padding-left: 0.5em;
}
div.imageblock div.content { padding-left: 0; }
span.image img { border-style: none; vertical-align: text-bottom; }
a.image:visited { color: white; }
dl {
margin-top: 0.8em;
margin-bottom: 0.8em;
}
dt {
margin-top: 0.5em;
margin-bottom: 0;
font-style: normal;
color: navy;
}
dd > *:first-child {
margin-top: 0.1em;
}
ul, ol {
list-style-position: outside;
}
ol.arabic {
list-style-type: decimal;
}
ol.loweralpha {
list-style-type: lower-alpha;
}
ol.upperalpha {
list-style-type: upper-alpha;
}
ol.lowerroman {
list-style-type: lower-roman;
}
ol.upperroman {
list-style-type: upper-roman;
}
div.compact ul, div.compact ol,
div.compact p, div.compact p,
div.compact div, div.compact div {
margin-top: 0.1em;
margin-bottom: 0.1em;
}
tfoot {
font-weight: bold;
}
td > div.verse {
white-space: pre;
}
div.hdlist {
margin-top: 0.8em;
margin-bottom: 0.8em;
}
div.hdlist tr {
padding-bottom: 15px;
}
dt.hdlist1.strong, td.hdlist1.strong {
font-weight: bold;
}
td.hdlist1 {
vertical-align: top;
font-style: normal;
padding-right: 0.8em;
color: navy;
}
td.hdlist2 {
vertical-align: top;
}
div.hdlist.compact tr {
margin: 0;
padding-bottom: 0;
}
.comment {
background: yellow;
}
.footnote, .footnoteref {
font-size: 0.8em;
}
span.footnote, span.footnoteref {
vertical-align: super;
}
#footnotes {
margin: 20px 0 20px 0;
padding: 7px 0 0 0;
}
#footnotes div.footnote {
margin: 0 0 5px 0;
}
#footnotes hr {
border: none;
border-top: 1px solid silver;
height: 1px;
text-align: left;
margin-left: 0;
width: 20%;
min-width: 100px;
}
div.colist td {
padding-right: 0.5em;
padding-bottom: 0.3em;
vertical-align: top;
}
div.colist td img {
margin-top: 0.3em;
}
@media print {
#footer-badges { display: none; }
}
#toc {
margin-bottom: 2.5em;
}
#toctitle {
color: #527bbd;
font-size: 1.1em;
font-weight: bold;
margin-top: 1.0em;
margin-bottom: 0.1em;
}
div.toclevel0, div.toclevel1, div.toclevel2, div.toclevel3, div.toclevel4 {
margin-top: 0;
margin-bottom: 0;
}
div.toclevel2 {
margin-left: 2em;
font-size: 0.9em;
}
div.toclevel3 {
margin-left: 4em;
font-size: 0.9em;
}
div.toclevel4 {
margin-left: 6em;
font-size: 0.9em;
}
span.aqua { color: aqua; }
span.black { color: black; }
span.blue { color: blue; }
span.fuchsia { color: fuchsia; }
span.gray { color: gray; }
span.green { color: green; }
span.lime { color: lime; }
span.maroon { color: maroon; }
span.navy { color: navy; }
span.olive { color: olive; }
span.purple { color: purple; }
span.red { color: red; }
span.silver { color: silver; }
span.teal { color: teal; }
span.white { color: white; }
span.yellow { color: yellow; }
span.aqua-background { background: aqua; }
span.black-background { background: black; }
span.blue-background { background: blue; }
span.fuchsia-background { background: fuchsia; }
span.gray-background { background: gray; }
span.green-background { background: green; }
span.lime-background { background: lime; }
span.maroon-background { background: maroon; }
span.navy-background { background: navy; }
span.olive-background { background: olive; }
span.purple-background { background: purple; }
span.red-background { background: red; }
span.silver-background { background: silver; }
span.teal-background { background: teal; }
span.white-background { background: white; }
span.yellow-background { background: yellow; }
span.big { font-size: 2em; }
span.small { font-size: 0.6em; }
span.underline { text-decoration: underline; }
span.overline { text-decoration: overline; }
span.line-through { text-decoration: line-through; }
div.unbreakable { page-break-inside: avoid; }
/*
* xhtml11 specific
*
* */
div.tableblock {
margin-top: 1.0em;
margin-bottom: 1.5em;
}
div.tableblock > table {
border: 3px solid #527bbd;
}
thead, p.table.header {
font-weight: bold;
color: #527bbd;
}
p.table {
margin-top: 0;
}
/* Because the table frame attribute is overriden by CSS in most browsers. */
div.tableblock > table[frame="void"] {
border-style: none;
}
div.tableblock > table[frame="hsides"] {
border-left-style: none;
border-right-style: none;
}
div.tableblock > table[frame="vsides"] {
border-top-style: none;
border-bottom-style: none;
}
/*
* html5 specific
*
* */
table.tableblock {
margin-top: 1.0em;
margin-bottom: 1.5em;
}
thead, p.tableblock.header {
font-weight: bold;
color: #527bbd;
}
p.tableblock {
margin-top: 0;
}
table.tableblock {
border-width: 3px;
border-spacing: 0px;
border-style: solid;
border-color: #527bbd;
border-collapse: collapse;
}
th.tableblock, td.tableblock {
border-width: 1px;
padding: 4px;
border-style: solid;
border-color: #527bbd;
}
table.tableblock.frame-topbot {
border-left-style: hidden;
border-right-style: hidden;
}
table.tableblock.frame-sides {
border-top-style: hidden;
border-bottom-style: hidden;
}
table.tableblock.frame-none {
border-style: hidden;
}
th.tableblock.halign-left, td.tableblock.halign-left {
text-align: left;
}
th.tableblock.halign-center, td.tableblock.halign-center {
text-align: center;
}
th.tableblock.halign-right, td.tableblock.halign-right {
text-align: right;
}
th.tableblock.valign-top, td.tableblock.valign-top {
vertical-align: top;
}
th.tableblock.valign-middle, td.tableblock.valign-middle {
vertical-align: middle;
}
th.tableblock.valign-bottom, td.tableblock.valign-bottom {
vertical-align: bottom;
}
/*
* manpage specific
*
* */
body.manpage h1 {
padding-top: 0.5em;
padding-bottom: 0.5em;
border-top: 2px solid silver;
border-bottom: 2px solid silver;
}
body.manpage h2 {
border-style: none;
}
body.manpage div.sectionbody {
margin-left: 3em;
}
@media print {
body.manpage div#toc { display: none; }
}
</style>
<script type="text/javascript">
/*<+'])');
// Function that scans the DOM tree for header elements (the DOM2
// nodeIterator API would be a better technique but not supported by all
// browsers).
var iterate = function (el) {
for (var i = el.firstChild; i != null; i = i.nextSibling) {
if (i.nodeType == 1 /* Node.ELEMENT_NODE */) {
var mo = re.exec(i.tagName);
if (mo && (i.getAttribute("class") || i.getAttribute("className")) != "float") {
result[result.length] = new TocEntry(i, getText(i), mo[1]-1);
}
iterate(i);
}
}
}
iterate(el);
return result;
}
var toc = document.getElementById("toc");
if (!toc) {
return;
}
// Delete existing TOC entries in case we're reloading the TOC.
var tocEntriesToRemove = [];
var i;
for (i = 0; i < toc.childNodes.length; i++) {
var entry = toc.childNodes[i];
if (entry.nodeName.toLowerCase() == 'div'
&& entry.getAttribute("class")
&& entry.getAttribute("class").match(/^toclevel/))
tocEntriesToRemove.push(entry);
}
for (i = 0; i < tocEntriesToRemove.length; i++) {
toc.removeChild(tocEntriesToRemove[i]);
}
// Rebuild TOC entries.
var entries = tocEntries(document.getElementById("content"), toclevels);
for (var i = 0; i < entries.length; ++i) {
var entry = entries[i];
if (entry.element.id == "")
entry.element.id = "_toc_" + i;
var a = document.createElement("a");
a.href = "#" + entry.element.id;
a.appendChild(document.createTextNode(entry.text));
var div = document.createElement("div");
div.appendChild(a);
div.className = "toclevel" + entry.toclevel;
toc.appendChild(div);
}
if (entries.length == 0)
toc.parentNode.removeChild(toc);
},
/////////////////////////////////////////////////////////////////////
// Footnotes generator
/////////////////////////////////////////////////////////////////////
/* Based on footnote generation code from:
* http://www.brandspankingnew.net/archive/2005/07/format_footnote.html
*/
footnotes: function () {
// Delete existing footnote entries in case we're reloading the footnodes.
var i;
var noteholder = document.getElementById("footnotes");
if (!noteholder) {
return;
}
var entriesToRemove = [];
for (i = 0; i < noteholder.childNodes.length; i++) {
var entry = noteholder.childNodes[i];
if (entry.nodeName.toLowerCase() == 'div' && entry.getAttribute("class") == "footnote")
entriesToRemove.push(entry);
}
for (i = 0; i < entriesToRemove.length; i++) {
noteholder.removeChild(entriesToRemove[i]);
}
// Rebuild footnote entries.
var cont = document.getElementById("content");
var spans = cont.getElementsByTagName("span");
var refs = {};
var n = 0;
for (i=0; i<spans.length; i++) {
if (spans[i].className == "footnote") {
n++;
var note = spans[i].getAttribute("data-note");
if (!note) {
// Use [\s\S] in place of . so multi-line matches work.
// Because JavaScript has no s (dotall) regex flag.
note = spans[i].innerHTML.match(/\s*\[([\s\S]*)]\s*/)[1];
spans[i].innerHTML =
"[<a id='_footnoteref_" + n + "' href='#_footnote_" + n +
"' title='View footnote' class='footnote'>" + n + "</a>]";
spans[i].setAttribute("data-note", note);
}
noteholder.innerHTML +=
"<div class='footnote' id='_footnote_" + n + "'>" +
"<a href='#_footnoteref_" + n + "' title='Return to text'>" +
n + "</a>. " + note + "</div>";
var id =spans[i].getAttribute("id");
if (id != null) refs["#"+id] = n;
}
}
if (n == 0)
noteholder.parentNode.removeChild(noteholder);
else {
// Process footnoterefs.
for (i=0; i<spans.length; i++) {
if (spans[i].className == "footnoteref") {
var href = spans[i].getElementsByTagName("a")[0].getAttribute("href");
href = href.match(/#.*/)[0]; // Because IE return full URL.
n = refs[href];
spans[i].innerHTML =
"[<a href='#_footnote_" + n +
"' title='View footnote' class='footnote'>" + n + "</a>]";
}
}
}
},
install: function(toclevels) {
var timerId;
function reinstall() {
asciidoc.footnotes();
if (toclevels) {
asciidoc.toc(toclevels);
}
}
function reinstallAndRemoveTimer() {
clearInterval(timerId);
reinstall();
}
timerId = setInterval(reinstall, 500);
if (document.addEventListener)
document.addEventListener("DOMContentLoaded", reinstallAndRemoveTimer, false);
else
window.onload = reinstallAndRemoveTimer;
}
}
asciidoc.install();
/*]]>*/
</script>
</head>
<body class="article">
<div id="header">
<h1>Converting Recoll indexing to multithreading</h1>
<span id="author">Jean-Fran��ois Dock��s</span><br />
<span id="email"><code><<a href="mailto:jfd@recoll.org">jfd@recoll.org</a>></code></span><br />
<span id="revdate">2012-12-03</span>
</div>
<div id="content">
<div class="sect1">
<h2 id="_abstract">Abstract</h2>
<div class="sectionbody">
<div class="paragraph"><p>This relates lessons learned while modifying <strong>Recoll</strong> indexing to be
multithreaded. I am by no means a threaded applications expert, so that a
few of the observations I made whole doing this may be of use to other
novices.</p></div>
</div>
</div>
<div class="sect1">
<h2 id="_introduction">Introduction</h2>
<div class="sectionbody">
<div class="paragraph"><p><a href="http://www.recoll.org"><strong>Recoll</strong></a> is a document indexing application, it
allows you to find documents by specifying search terms.</p></div>
<div class="paragraph"><p>The documents need to be <em>indexed</em> for searches to be fast. In a nutshell,
we convert the different document formats to text, then split the text into
terms and remember where those occur. This is a time-consuming operation.</p></div>
<div class="paragraph"><p>Up to version 1.18 <strong>Recoll</strong> indexing is single-threaded: routines which
call each other sequentially.</p></div>
<div class="paragraph"><p>In most personal indexer contexts, it is also CPU-bound. There is a lot of
conversion work necessary for turning those PDF (or other) files into
appropriately cleaned up pure text, then split it into terms and update the
index. Given the relatively modest amount of data, and the speed of
storage, I/O issues are secondary.</p></div>
<div class="paragraph"><p>Looking at the <em>CPU idle</em> <strong>top</strong> output stuck at 75% on my quad-core CPU,
while waiting for the indexing to finish, was frustrating, and I was
tempted to find a way to keep those other cores at temperature and shorten
the waiting.</p></div>
<div class="paragraph"><p>For some usages, the best way to accomplish this may be to just partition
the index and independantly start indexing on different configurations,
using multiple processes to better utilize the available processing power.</p></div>
<div class="paragraph"><p>This is not an universal solution though, as it is complicated to set up,
not optimal in general for indexing performance, and not always optimal
either at query time.</p></div>
<div class="paragraph"><p>The most natural way to improve indexing times is to increase CPU
utilization by using multiple threads inside an indexing process.</p></div>
<div class="paragraph"><p>Something similar had been done with earlier versions of the <strong>Recoll</strong> GUI,
which had an internal indexing thread. This had been a frequent source of
trouble though, and linking the GUI and indexing process lifetimes was a
bad idea, so, in recent versions, the indexing is always performed by an
external process. Still, this experience had put in light most of the
problem areas, and prepared the code for further work.</p></div>
<div class="paragraph"><p>It should be noted that, as <code>recollindex</code> is both <em>nice</em>'d and <em>ionice</em>'d
as a lowest priority process, it will only use free computing power on the
machine, and will step down as soon as anything else wants to work.</p></div>
<div class="sidebarblock">
<div class="content">
<div class="paragraph"><p>The only case where you may notice that the indexing is at work
is when the machine is short on memory and things (such as
your Web browser) get swapped-out while you are not actively using
them. You then notice a long delay when you want to start, because they
need to be swapped back in. There is little which can be done about
this. Setting <em>idxflushmb</em> to a low value may help in some cases (depending
on the document sizes). May I also suggest in this case that, if your
machine can take more memory, it may be a good idea to procure some, as
memory is nowadays quite cheap, and memory-starved machines are not fun.</p></div>
</div></div>
<div class="paragraph"><p>In general, augmenting the machine utilisation by <code>recollindex</code> just does
not change its responsiveness. My PC has a an Intel Pentium Core i5 750 (4
cores, no hyperthreading), which is far from being a high performance CPU
(nowadays…), and I often forget that I am running indexing tests, it is
just not noticeable. The machine does have a lot of memory though (12GB).</p></div>
</div>
</div>
<div class="sect1">
<h2 id="_the_recoll_indexing_processing_flow">The Recoll indexing processing flow</h2>
<div class="sectionbody">
<div class="imageblock" style="float:right;">
<div class="content">
<img src="nothreads.png" alt="Basic flow" />
</div>
</div>
<div class="paragraph"><p>There are 4 main steps in the <code>recollindex</code> processing pipeline:</p></div>
<div class="olist arabic"><ol class="arabic">
<li>
<p>
Find the file
</p>
</li>
<li>
<p>
Convert it to text
</p>
</li>
<li>
<p>
Process the text (split, strip etc.) and create a <strong>Xapian</strong> document
</p>
</li>
<li>
<p>
Update the index
</p>
</li>
</ol></div>
<div class="paragraph"><p>The first step, walking the file system (or some other data source), is
usually much faster than the others, and we just leave it alone to be
performed by the main thread. It outputs file names (and the associated
<strong>POSIX</strong> <em>stat</em> data).</p></div>
<div class="paragraph"><p>The last step, <strong>Xapian</strong> index updating, can only be single-threaded.</p></div>
<div class="paragraph"><p>The first idea is to change the indexing pipeline so that each step is
performed by an independant worker thread, passing its output to the next
thread, in assembly-line fashion.</p></div>
<div class="paragraph"><p>In order to achieve this, we need to decouple the different phases. They
are normally linked by procedure calls, which we replace with a job
control object: the <em>WorkQueue</em>.</p></div>
<div class="sect2">
<h3 id="_the_workqueue">The WorkQueue</h3>
<div class="paragraph"><p>The <em>WorkQueue</em> object is implemented by a reasonably simple class, which
manages an input queue on which client append jobs, and a set of worker
threads, which retrieve and perform the jobs, and whose lifetime are
managed by the <em>WorkQueue</em> object. The
<a href="https://bitbucket.org/medoc/recoll/src/f06f3aba912045d6ad52e9a0fd930b95e363fd10/src/utils/workqueue.h?at=default">implementation</a> is straightforward with
<strong>POSIX</strong> threads synchronization functions and C++ <strong>STL</strong> data structures.</p></div>
<div class="paragraph"><p>In practise it proved quite simple to modify existing code to create a job
object and put it on the queue, instead of calling the downstream routine
with the job parameters, <em>while keeping the capacity to call the downstream
routine directly</em>. The kind of coupling is determined either by compilation
flags (for global disabling/enabling of multithreading), or according to
configuration data, which allows experimenting with different threads
arrangements just by changing parameters in a file, without recompiling.</p></div>
<div class="paragraph"><p>Each <em>WorkQueue</em> accepts two parameters: the length of the input queue
(before a client will block when trying to add a job), and the number of
worker threads. Both parameters can be set in the <strong>Recoll</strong> configuration
file for each of the three queues used in the indexing pipeline. Setting
the queue length to -1 will disable the corresponding queue (using a direct
call instead).</p></div>
<div style="clear:both;"></div>
</div>
</div>
</div>
<div class="sect1">
<h2 id="_the_assembly_line">The Assembly Line</h2>
<div class="sectionbody">
<div class="imageblock" style="float:right;">
<div class="content">
<img src="assembly.png" alt="Assembly line" />
</div>
</div>
<div class="paragraph"><p>So the first idea is to create 3 explicit threads to manage the file
conversion, the term generation, and the <strong>Xapian</strong> index update. The first
thread prepares a file, passes it on to the term generation thread, and
immediately goes back to work on the next file, etc.</p></div>
<div class="paragraph"><p>The presumed advantage of this method is that the different stages, which
perform disjointed processing, should share little, so that we can hope to
minimize the changes necessitated by the threads interactions.</p></div>
<div class="paragraph"><p>However some changes to the code were needed to make this work (and a few
bugs were missed, which only became apparent at later stages, confirming
that the <em>low interaction</em> idea was not completely false).</p></div>
<div class="sect2">
<h3 id="_converting_to_multithreading_what_to_look_for">Converting to multithreading: what to look for</h3>
<div class="paragraph"><p>I am probably stating the obvious here, but when preparing a program for
multi-threading, problems can only arise where non-constant data is
accessed by different threads.</p></div>
<div class="paragraph"><p>Once you have solved the core problems posed by the obvious data that needs
to be shared, you will be left to deal with less obvious, hidden,
interactions inside the program.</p></div>
<div class="paragraph"><p>Classically this would concern global or static data, but in a C++ program,
class members will be a concern if a single object can be accessed by
several threads.</p></div>
<div class="paragraph"><p>Hunting for static data inside a program of non trivial size is not always
obvious. Two approaches can be used: hunting for the <em>static</em> keyword in
source code, or looking at global and static data symbols in <strong>nm</strong> output.</p></div>
<div class="paragraph"><p>Once found, there are mostly three types of static/global data:</p></div>
<div class="ulist"><ul>
<li>
<p>
Things that need to be eliminated: for example, routines can be made
reentrant by letting the caller supply a storage buffer instead of using
an internal static one (which was a bad idea in the first place
anyway).
</p>
</li>
<li>
<p>
Things that need to be protected: sometimes, the best approach is just
to protect the access with a mutex lock. It is trivial to encapsulate
the locks in C++ objects to use the "Resource Acquisition is
Initialization" idiom, easily making sure that locks are freed when
exiting the critical section. A very basic
<a href="https://bitbucket.org/medoc/recoll/src/f06f3aba9120/src/utils/ptmutex.h?at=default">example of implementation</a>
can be found in the <strong>Recoll</strong> source code.
</p>
</li>
<li>
<p>
Things which can stay: this is mostly initialization data such as value
tables which are computed once, and then stay logically constant during
program execution. In order to be sure of a correct single-threaded
initialization, it is best to explicitly initialize the modules or
functions that use this kind of data in the main thread when the program
starts.
</p>
</li>
</ul></div>
</div>
<div class="sect2">
<h3 id="_assembly_line_approach_the_results">Assembly line approach: the results</h3>
<div class="paragraph"><p>Unfortunately, the assembly line approach yields very modest improvements
when used inside <strong>Recoll</strong> indexing. The reason, is that this method needs
stages of equivalent complexity to be efficient. If one of the stages
dominates the others, its thread will be the only one active at any time,
and little will be gained.</p></div>
<div class="paragraph"><p>What is especially problematic is that the balance between tasks need not
only exist on average, but also for the majority of individual jobs.</p></div>
<div class="paragraph"><p>For <strong>Recoll</strong> indexing, even if the data preparation and index update steps
are often of the same order of magnitude <em>on average</em>, their balance
depends a lot on the kind of data being processed, so that things are
usually unbalanced at any given time: the index update thread is mostly
idle while processing PDF files, and the data preparation has little to do
when working on HTML or plain text.</p></div>
<div class="paragraph"><p>In practice, very modest indexing time improvements from 5% to 15% were
achieved with this method.</p></div>
</div>
</div>
</div>
<div class="sect1">
<h2 id="recoll.idxthreads.multistage">The next step: multi-stage parallelism</h2>
<div class="sectionbody">
<div class="imageblock" style="float:right;">
<div class="content">
<img src="multipara.png" alt="Multi-stage parallelism" />
</div>
</div>
<div class="paragraph"><p>Given the limitations of the assembly line approach, the next step in the
transformation of <strong>Recoll</strong> indexing was to enable full parallelism wherever
possible.</p></div>
<div class="paragraph"><p>Of the four processing steps (see figures), two are not candidates for
parallelization:</p></div>
<div class="ulist"><ul>
<li>
<p>
File system walking is so fast compared to the other steps that using
several threads would make no sense (it would also quite probably become
IO bound if we tried anyway).
</p>
</li>
<li>
<p>
The <strong>Xapian</strong> library index updating code is not designed for
multi-threading and must stay protected from multiple accesses.
</p>
</li>
</ul></div>
<div class="paragraph"><p>The two other steps are good candidates.</p></div>
<div class="paragraph"><p>Most of the work to make <strong>Recoll</strong> code reentrant had been performed for the
previous transformation. Going full-parallel only implied protecting the
data structures that needed to be shared by the threads performing a given
processing step.</p></div>
<div class="paragraph"><p>Just for the anecdotic value, a list of the elements that needed mutexes:</p></div>
<div class="ulist"><ul>
<li>
<p>
Filter subprocesses cache: some file conversion subprocesses may be
expensive (starting a Python process is no piece of cake), so they are
cached for reuse after they are done translating a file. The shared cache
needs protection.
</p>
</li>
<li>
<p>
Status updates: an object used to update the current file name and indexing
status to a shared file.
</p>
</li>
<li>
<p>
Missing store: the list of missing helper programs
</p>
</li>
<li>
<p>
The readonly <strong>Xapian</strong> database object: a Xapian::Database object which is
used for checking the validity of current index data against a file’s
last modification date.
</p>
</li>
<li>
<p>
Document existence map: a bit array used to store an existence bit about
every document, and purge the disappeared at the end of the indexing
pass. This is accessed both from the file conversion and database update
code, so it also needed protection in the previous assembly line
approach.
</p>
</li>
<li>
<p>
Mbox offsets cache. Used to store the offsets of individual messages
inside <strong>mbox</strong> files.
</p>
</li>
<li>
<p>
<strong>iconv</strong> control blocks: these are cached for reuse in several places, and
need protection. Actually, it might be better in multithreading context
to just suppress the reuse and locking. Rough tests seem to indicate that
the impact on overall performance is small, but this might change with
higher parallelism (or not…).
</p>
</li>
</ul></div>
<div class="paragraph"><p>The <strong>Recoll</strong> configuration also used to be managed by a single shared
object, which is mutable as values may depend on what area of the
file-system we are exploring, so that the object is stateful and updated as
we change directories. The choice made here was to duplicate the object
where needed (each indexing thread gets its own). This gave rise to the
sneakiest bug in the whole transformation (see further down).</p></div>
<div class="paragraph"><p>Having a dynamic way to define the threads configuration makes it easy to
experiment. For example, the following data defines the configuration that
was finally found to be best overall on my hardware:</p></div>
<div class="literalblock">
<div class="content">
<pre><code>thrQSizes = 2 2 2
thrTCounts = 4 2 1</code></pre>
</div></div>
<div class="paragraph"><p>This is using 3 queues of depth 2, 4 threads working on file conversion, 2
on text splitting and other document processing, and 1 on Xapian updating
(no choice here).</p></div>
<div style="clear:both;"></div>
</div>
</div>
<div class="sect1">
<h2 id="_bench_results">Bench results</h2>
<div class="sectionbody">
<div class="paragraph"><p>So the big question after all the work: was it worth it ? I could only get
a real answer when the program stopped crashing, so this took some time and
a little faith, but the answer is positive, as far as I’m
concerned. Performance has improved significantly and this was a fun
project.</p></div>
<div class="tableblock">
<table rules="all"
width="70%"
frame="border"
cellspacing="0" cellpadding="4">
<caption class="title">Table 1. Results on a variety of file system areas:</caption>
<col width="20%" />
<col width="20%" />
<col width="20%" />
<col width="20%" />
<col width="20%" />
<thead>
<tr>
<th align="left" valign="top">Area </th>
<th align="left" valign="top">Seconds before </th>
<th align="left" valign="top">Seconds after</th>
<th align="left" valign="top"> Percent Improvement</th>
<th align="left" valign="top"> Speed Factor</th>
</tr>
</thead>
<tbody>
<tr>
<td align="left" valign="top"><p class="table">home</p></td>
<td align="left" valign="top"><p class="table">12742</p></td>
<td align="left" valign="top"><p class="table">6942</p></td>
<td align="left" valign="top"><p class="table">46%</p></td>
<td align="left" valign="top"><p class="table">1.8</p></td>
</tr>
<tr>
<td align="left" valign="top"><p class="table">mail</p></td>
<td align="left" valign="top"><p class="table">2700</p></td>
<td align="left" valign="top"><p class="table">1563</p></td>
<td align="left" valign="top"><p class="table">58%</p></td>
<td align="left" valign="top"><p class="table">1.7</p></td>
</tr>
<tr>
<td align="left" valign="top"><p class="table">projets</p></td>
<td align="left" valign="top"><p class="table">5022</p></td>
<td align="left" valign="top"><p class="table">1970</p></td>
<td align="left" valign="top"><p class="table">61%</p></td>
<td align="left" valign="top"><p class="table">2.5</p></td>
</tr>
<tr>
<td align="left" valign="top"><p class="table">pdf</p></td>
<td align="left" valign="top"><p class="table">2164</p></td>
<td align="left" valign="top"><p class="table">770</p></td>
<td align="left" valign="top"><p class="table">64%</p></td>
<td align="left" valign="top"><p class="table">2.8</p></td>
</tr>
<tr>
<td align="left" valign="top"><p class="table">otherhtml</p></td>
<td align="left" valign="top"><p class="table">5593</p></td>
<td align="left" valign="top"><p class="table">4014</p></td>
<td align="left" valign="top"><p class="table">28%</p></td>
<td align="left" valign="top"><p class="table">1.4</p></td>
</tr>
</tbody>
</table>
</div>
<div class="tableblock">
<table rules="all"
width="70%"
frame="border"
cellspacing="0" cellpadding="4">
<caption class="title">Table 2. Characteristics of the data</caption>
<col width="20%" />
<col width="20%" />
<col width="20%" />
<col width="20%" />
<col width="20%" />
<thead>
<tr>
<th align="left" valign="top">Area </th>
<th align="left" valign="top"> Files MB </th>
<th align="left" valign="top"> Files </th>
<th align="left" valign="top"> DB MB </th>
<th align="left" valign="top"> Documents</th>
</tr>
</thead>
<tbody>
<tr>
<td align="left" valign="top"><p class="table">home</p></td>
<td align="left" valign="top"><p class="table">64106</p></td>
<td align="left" valign="top"><p class="table">44897</p></td>
<td align="left" valign="top"><p class="table">1197</p></td>
<td align="left" valign="top"><p class="table">104797</p></td>
</tr>
<tr>
<td align="left" valign="top"><p class="table">mail</p></td>
<td align="left" valign="top"><p class="table">813</p></td>
<td align="left" valign="top"><p class="table">232</p></td>
<td align="left" valign="top"><p class="table">663</p></td>
<td align="left" valign="top"><p class="table">47267</p></td>
</tr>
<tr>
<td align="left" valign="top"><p class="table">projets</p></td>
<td align="left" valign="top"><p class="table">2056</p></td>
<td align="left" valign="top"><p class="table">34504</p></td>
<td align="left" valign="top"><p class="table">549</p></td>
<td align="left" valign="top"><p class="table">40281</p></td>
</tr>
<tr>
<td align="left" valign="top"><p class="table">pdf</p></td>
<td align="left" valign="top"><p class="table">1123</p></td>
<td align="left" valign="top"><p class="table">1139</p></td>
<td align="left" valign="top"><p class="table">111</p></td>
<td align="left" valign="top"><p class="table">1139</p></td>
</tr>
<tr>
<td align="left" valign="top"><p class="table">otherhtml</p></td>
<td align="left" valign="top"><p class="table">3442</p></td>
<td align="left" valign="top"><p class="table">223007</p></td>
<td align="left" valign="top"><p class="table">2080</p></td>
<td align="left" valign="top"><p class="table">221890</p></td>
</tr>
</tbody>
</table>
</div>
<div class="paragraph"><p><em>home</em> is my home directory. The high megabyte value is due to a number of
very big and not indexed <strong>VirtualBox</strong> images. Otherwise, it’s a wide
mix of source files, email, miscellaneous documents and ebooks.</p></div>
<div class="paragraph"><p><em>mail</em> is my mail directory, full of <strong>mbox</strong> files.</p></div>
<div class="paragraph"><p><em>projets</em> mostly holds source files, and a number of documents.</p></div>
<div class="paragraph"><p><em>pdf</em> holds random <strong>pdf</strong> files harvested on the internets. The performance
is quite spectacular, because most of the processing time goes to
converting them to text, and this is done in parallel. Probably could be
made a bit faster with more cores, until we hit the <strong>Xapian</strong> update speed
limit.</p></div>
<div class="paragraph"><p><em>otherhtml</em> holds myriad of small html files, mostly from
<strong>wikipedia</strong>. The improvement is not great here because a lot of time is
spent in the single-threaded <strong>Xapian</strong> index update.</p></div>
<div class="paragraph"><p>The tests were made with queue depths of 2 on all queues, and 4 threads
working on the file conversion step, 2 on the term generation.</p></div>
</div>
</div>
<div class="sect1">
<h2 id="_a_variation_linear_parallelism">A variation: linear parallelism</h2>
<div class="sectionbody">
<div class="paragraph"><p>Once past the assembly-line idea, another possible transformation would be
to get rid of the two downstream queues, and just create a job for each
file and let it go to the end (using a mutex to protect accesses to the
writable <strong>Xapian</strong> database).</p></div>
<div class="paragraph"><p>With the current <strong>Recoll</strong> code, this can be defined by the following
parameters (one can also use a deeper front queue, this changes little):</p></div>
<div class="literalblock">
<div class="content">
<pre><code>thrQSizes = 2 -1 -1
thrTCounts = 4 0 0</code></pre>
</div></div>
<div class="paragraph"><p>In practise, the performance is close to the one for the multistage
version.</p></div>
<div class="paragraph"><p>If we were to hard-code this approach, this would be a simpler
modification, necessitating less changes to the code, but it has a slight
inconvenient: when working on a single big multi-document file, no
parallelism at all can be obtained. In this situation, the multi-stage
approach brings us back to the assembly-line behaviour, so the improvements
are not great, but they do exist.</p></div>
</div>
</div>
<div class="sect1">
<h2 id="_miscellany">Miscellany</h2>
<div class="sectionbody">
<div class="sect2">
<h3 id="_the_big_gotcha_my_stack_dump_staring_days">The big gotcha: my stack dump staring days</h3>
<div class="paragraph"><p>Overall, debugging the modified program was reasonably
straightforward. Data access synchronization issues mostly provoke dynamic
data corruption, which can be beastly to debug. I was lucky enough that
most crashes occurred in the code that was actually related to the
corrupted data, not in some randomly located and unrelated dynamic memory
user, so that the issues were reasonably easy to find.</p></div>
<div class="paragraph"><p>One issue though kept me working for a few days. The indexing process kept
crashing randomly at an interval of a few thousands documents, segfaulting
on a bad pointer. An access to the configuration data structure seemed to
be involved, but, as each thread was supposed to have its own copy, I was
out of ideas.</p></div>
<div class="paragraph"><p>After reviewing all the uses for the configuration data (there are quite a
few), the problem was finally revealed to lie with the filter process
cache. Each filter structure stored in the cache stores a pointer to a
configuration structure. This belonged to the thread which initially
created the filter. But the filter would often be reused by a different
thread, with the consequence that the configuration object was now accessed
and modified by two unsynchronized threads… Resetting the config pointer
at the time of filter reuse was the
<a href="https://bitbucket.org/medoc/recoll/commits/943de4b78818079b0eb6ffd0fcbdfdd0746b4a40">ridiculously
simple (almost)single-line fix</a> to this evasive problem.</p></div>
<div class="paragraph"><p>Looking at multi-threaded stack dumps is mostly fun for people with several
heads, which is unfortunately not my case, so I was quite elated when this
was over.</p></div>
</div>
<div class="sect2">
<h3 id="_fork_performance_issues">Fork performance issues</h3>
<div class="paragraph"><p>On a quite unrelated note, something that I discovered while evaluating the
program performance is that forking a big process like <code>recollindex</code> can be
quite expensive. Even if the memory space of the forked process is not
copied (it’s Copy On Write, and we write very little before the following
exec), just duplicating the memory maps can be slow when the process uses a
few hundred megabytes.</p></div>
<div class="paragraph"><p>I modified the single-threaded version of <code>recollindex</code> to use <strong>vfork</strong>
instead of <strong>fork</strong>, but this can’t be used with multiple threads (no
modification of the process memory space is allowed in the child between
<strong>vfork</strong> and <strong>exec</strong>, so we’d have to have a way to suspend all the threads
first).</p></div>
<div class="paragraph"><p>I did not implement a solution to this issue, and I don’t think
that a simple one exists. The workaround is to use modest <strong>Xapian</strong> flush
values to prevent the process from becoming too big.</p></div>
<div class="paragraph"><p>A longer time solution would be to implement a small slave process to do
the executing of ephemeral external commands.</p></div>
</div>
</div>
</div>
</div>
<div id="footnotes"><hr /></div>
<div id="footer">
<div id="footer-text">
Last updated
2016-08-07 15:42:01 CEST
</div>
</div>
</body>
</html>