1 <?xml version="1.0" encoding="UTF-8"?>
\r
2 <chapter xml:id="opensrf" xmlns="http://docbook.org/ns/docbook" version="5.0" xml:lang="EN"
\r
3 xmlns:xi="http://www.w3.org/2001/XInclude" xmlns:xlink="http://www.w3.org/1999/xlink">
\r
5 <title>OpenSRF</title>
\r
7 <abstract id="openSRF_abstract">
\r
8 <simpara>One of the claimed advantages of
\r
9 Evergreen over alternative integrated library systems is the underlying Open
\r
10 Service Request Framework (OpenSRF, pronounced "open surf") architecture. This
\r
11 article introduces OpenSRF, demonstrates how to build OpenSRF services through
\r
12 simple code examples, and explains the technical foundations on which OpenSRF
\r
13 is built. This chapter was taken from Dan Scott's <emphasis>Easing gently into OpenSRF</emphasis> article, June, 2010.</simpara>
\r
15 <section id="_introducing_opensrf">
\r
16 <title>Introducing OpenSRF</title>
\r
17 <simpara>OpenSRF is a message routing network that offers scalability and failover
\r
18 support for individual services and entire servers with minimal development and
\r
19 deployment overhead. You can use OpenSRF to build loosely-coupled applications
\r
20 that can be deployed on a single server or on clusters of geographically
\r
21 distributed servers using the same code and minimal configuration changes.
\r
22 Although copyright statements on some of the OpenSRF code date back to Mike
\r
23 Rylander’s original explorations in 2000, Evergreen was the first major
\r
24 application to be developed with, and to take full advantage of, the OpenSRF
\r
25 architecture starting in 2004. The first official release of OpenSRF was 0.1 in
\r
26 February 2005 (<ulink url="http://evergreen-ils.org/blog/?p=21">http://evergreen-ils.org/blog/?p=21</ulink>), but OpenSRF’s development
\r
27 continues a steady pace of enhancement and refinement, with the release of
\r
28 1.0.0 in October 2008 and the most recent release of 1.2.2 in February 2010.</simpara>
\r
29 <simpara>OpenSRF is a distinct break from the architectural approach used by previous
\r
30 library systems and has more in common with modern Web applications. The
\r
31 traditional "scale-up" approach to serve more transactions is to purchase a
\r
32 server with more CPUs and more RAM, possibly splitting the load between a Web
\r
33 server, a database server, and a business logic server. Evergreen, however, is
\r
34 built on the Open Service Request Framework (OpenSRF) architecture, which
\r
35 firmly embraces the "scale-out" approach of spreading transaction load over
\r
36 cheap commodity servers. The <ulink url="http://evergreen-ils.org/blog/?p=56">initial GPLS
\r
37 PINES hardware cluster</ulink>, while certainly impressive, may have offered the
\r
38 misleading impression that Evergreen requires a lot of hardware to run.
\r
39 However, Evergreen and OpenSRF easily scale down to a single server; many
\r
40 Evergreen libraries run their entire library system on a single server, and
\r
41 most OpenSRF and Evergreen development occurs on a virtual machine running on a
\r
42 single laptop or desktop image.</simpara>
\r
43 <simpara>Another common concern is that the flexibility of OpenSRF’s distributed
\r
44 architecture makes it complex to configure and to write new applications. This
\r
45 article demonstrates that OpenSRF itself is an extremely simple architecture on
\r
46 which one can easily build applications of many kinds – not just library
\r
47 applications – and that you can use a number of different languages to call and
\r
48 implement OpenSRF methods with a minimal learning curve. With an application
\r
49 built on OpenSRF, when you identify a bottleneck in your application’s business
\r
50 logic layer, you can adjust the number of the processes serving that particular
\r
51 bottleneck on each of your servers; or if the problem is that your service is
\r
52 resource-hungry, you could add an inexpensive server to your cluster and
\r
53 dedicate it to running that resource-hungry service.</simpara>
\r
54 <simplesect id="_programming_language_support">
\r
55 <title>Programming language support</title>
\r
56 <simpara>If you need to develop an entirely new OpenSRF service, you can choose from a
\r
57 number of different languages in which to implement that service. OpenSRF
\r
58 client language bindings have been written for C, Java, JavaScript, Perl, and
\r
59 Python, and service language bindings have been written for C, Perl, and Python.
\r
60 This article uses Perl examples as a lowest common denominator programming
\r
61 language. Writing an OpenSRF binding for another language is a relatively small
\r
62 task if that language offers libraries that support the core technologies on
\r
63 which OpenSRF depends:</simpara>
\r
67 <ulink url="http://tools.ietf.org/html/rfc3920">Extensible Messaging and Presence
\r
68 Protocol</ulink> (XMPP, sometimes referred to as Jabber) - provides the base messaging
\r
69 infrastructure between OpenSRF clients and services
\r
74 <ulink url="http://json.org">JavaScript Object Notation</ulink> (JSON) - serializes the content
\r
75 of each XMPP message in a standardized and concise format
\r
80 <ulink url="http://memcached.org">memcached</ulink> - provides the caching service
\r
85 <ulink url="http://tools.ietf.org/html/rfc5424">syslog</ulink> - the standard UNIX logging
\r
90 <simpara>Unfortunately, the
\r
91 <ulink url="http://evergreen-ils.org/dokuwiki/doku.php?id=osrf-devel:primer">OpenSRF
\r
92 reference documentation</ulink>, although augmented by the
\r
93 <ulink url="http://evergreen-ils.org/dokuwiki/doku.php?id=osrf-devel:terms">OpenSRF
\r
94 glossary</ulink>, blog posts like <ulink url="http://evergreen-ils.org/blog/?p=36">the description
\r
95 of OpenSRF and Jabber</ulink>, and even this article, is not a sufficient substitute
\r
96 for a complete specification on which one could implement a language binding.
\r
97 The recommended option for would-be developers of another language binding is
\r
98 to use the Python implementation as the cleanest basis for a port to another
\r
102 <section id="writing_an_opensrf_service">
\r
103 <title>Writing an OpenSRF Service</title>
\r
104 <simpara>Imagine an application architecture in which 10 lines of Perl or Python, using
\r
105 the data types native to each language, are enough to implement a method that
\r
106 can then be deployed and invoked seamlessly across hundreds of servers. You
\r
107 have just imagined developing with OpenSRF – it is truly that simple. Under the
\r
108 covers, of course, the OpenSRF language bindings do an incredible amount of
\r
109 work on behalf of the developer. An OpenSRF application consists of one or more
\r
110 OpenSRF services that expose methods: for example, the <literal>opensrf.simple-text</literal>
\r
111 <ulink url="http://svn.open-ils.org/trac/OpenSRF/browser/trunk/src/perl/lib/OpenSRF/Application/Demo/SimpleText.pm">demonstration
\r
112 service</ulink> exposes the <literal>opensrf.simple-text.split()</literal> and
\r
113 <literal>opensrf.simple-text.reverse()</literal> methods. Each method accepts zero or more
\r
114 arguments and returns zero or one results. The data types supported by OpenSRF
\r
115 arguments and results are typical core language data types: strings, numbers,
\r
116 booleans, arrays, and hashes.</simpara>
\r
117 <simpara>To implement a new OpenSRF service, perform the following steps:</simpara>
\r
118 <orderedlist numeration="arabic">
\r
121 Include the base OpenSRF support libraries
\r
126 Write the code for each of your OpenSRF methods as separate procedures
\r
131 Register each method
\r
136 Add the service definition to the OpenSRF configuration files
\r
140 <simpara>For example, the following code implements an OpenSRF service. The service
\r
141 includes one method named <literal>opensrf.simple-text.reverse()</literal> that accepts one
\r
142 string as input and returns the reversed version of that string:</simpara>
\r
143 <programlisting language="perl" linenumbering="unnumbered">
\r
146 package OpenSRF::Application::Demo::SimpleText;
\r
150 use OpenSRF::Application;
\r
151 use parent qw/OpenSRF::Application/;
\r
154 my ($self , $conn, $text) = @_;
\r
155 my $reversed_text = scalar reverse($text);
\r
156 return $reversed_text;
\r
159 __PACKAGE__->register_method(
\r
160 method => 'text_reverse',
\r
161 api_name => 'opensrf.simple-text.reverse'
\r
164 <simpara>Ten lines of code, and we have a complete OpenSRF service that exposes a single
\r
165 method and could be deployed quickly on a cluster of servers to meet your
\r
166 application’s ravenous demand for reversed strings! If you’re unfamiliar with
\r
167 Perl, the <literal>use OpenSRF::Application; use parent qw/OpenSRF::Application/;</literal>
\r
168 lines tell this package to inherit methods and properties from the
\r
169 <literal>OpenSRF::Application</literal> module. For example, the call to
\r
170 <literal>__PACKAGE__->register_method()</literal> is defined in <literal>OpenSRF::Application</literal> but due to
\r
171 inheritance is available in this package (named by the special Perl symbol
\r
172 <literal>__PACKAGE__</literal> that contains the current package name). The <literal>register_method()</literal>
\r
173 procedure is how we introduce a method to the rest of the OpenSRF world.</simpara>
\r
174 <simplesect id="serviceRegistration">
\r
175 <title>Registering a service with the OpenSRF configuration files</title>
\r
176 <simpara>Two files control most of the configuration for OpenSRF:</simpara>
\r
180 <literal>opensrf.xml</literal> contains the configuration for the service itself, as well as
\r
181 a list of which application servers in your OpenSRF cluster should start
\r
187 <literal>opensrf_core.xml</literal> (often referred to as the "bootstrap configuration"
\r
188 file) contains the OpenSRF networking information, including the XMPP server
\r
189 connection credentials for the public and private routers. You only need to touch
\r
190 this for a new service if the new service needs to be accessible via the
\r
195 <simpara>Begin by defining the service itself in <literal>opensrf.xml</literal>. To register the
\r
196 <literal>opensrf.simple-text</literal> service, add the following section to the <literal><apps></literal>
\r
197 element (corresponding to the XPath <literal>/opensrf/default/apps/</literal>):</simpara>
\r
198 <programlisting language="xml" linenumbering="unnumbered">
\r
200 <opensrf.simple-text> <co id="CO1-1"/>
\r
201 <keepalive>3</keepalive><co id="CO1-2"/>
\r
202 <stateless>1</stateless><co id="CO1-3"/>
\r
203 <language>perl</language><co id="CO1-4"/>
\r
204 <implementation>OpenSRF::Application::Demo::SimpleText</implementation><co id="CO1-5"/>
\r
205 <max_requests>100</max_requests><co id="CO1-6"/>
\r
206 <unix_config>
\r
207 <max_requests>1000</max_requests> <co id="CO1-7"/>
\r
208 <unix_log>opensrf.simple-text_unix.log</unix_log> <co id="CO1-8"/>
\r
209 <unix_sock>opensrf.simple-text_unix.sock</unix_sock><co id="CO1-9"/>
\r
210 <unix_pid>opensrf.simple-text_unix.pid</unix_pid> <co id="CO1-10"/>
\r
211 <min_children>5</min_children> <co id="CO1-11"/>
\r
212 <max_children>15</max_children><co id="CO1-12"/>
\r
213 <min_spare_children>2</min_spare_children><co id="CO1-13"/>
\r
214 <max_spare_children>5</max_spare_children> <co id="CO1-14"/>
\r
215 </unix_config>
\r
216 </opensrf.simple-text>
\r
218 <!-- other OpenSRF services registered here... -->
\r
222 <callout arearefs="CO1-1">
\r
224 The element name is the name that the OpenSRF control scripts use to refer
\r
228 <callout arearefs="CO1-2">
\r
230 The <literal><keepalive></literal> element specifies the interval (in seconds) between
\r
231 checks to determine if the service is still running.
\r
234 <callout arearefs="CO1-3">
\r
236 The <literal><stateless></literal> element specifies whether OpenSRF clients can call
\r
237 methods from this service without first having to create a connection to a
\r
238 specific service backend process for that service. If the value is <literal>1</literal>, then
\r
239 the client can simply issue a request and the router will forward the request
\r
240 to an available service and the result will be returned directly to the client.
\r
243 <callout arearefs="CO1-4">
\r
245 The <literal><language></literal> element specifies the programming language in which the
\r
246 service is implemented.
\r
249 <callout arearefs="CO1-5">
\r
251 The <literal><implementation></literal> element pecifies the name of the library or module
\r
252 in which the service is implemented.
\r
255 <callout arearefs="CO1-6">
\r
257 (C implementations only): The <literal><max_requests></literal> element, as a direct child
\r
258 of the service element name, specifies the maximum number of requests a process
\r
259 serves before it is killed and replaced by a new process.
\r
262 <callout arearefs="CO1-7">
\r
264 (Perl implementations only): The <literal><max_requests></literal> element, as a direct
\r
265 child of the <literal><unix_config></literal> element, specifies the maximum number of requests
\r
266 a process serves before it is killed and replaced by a new process.
\r
269 <callout arearefs="CO1-8">
\r
271 The <literal><unix_log></literal> element specifies the name of the log file for
\r
272 language-specific log messages such as syntax warnings.
\r
275 <callout arearefs="CO1-9">
\r
277 The <literal><unix_sock></literal> element specifies the name of the UNIX socket used for
\r
278 inter-process communications.
\r
281 <callout arearefs="CO1-10">
\r
283 The <literal><unix_pid></literal> element specifies the name of the PID file for the
\r
284 master process for the service.
\r
287 <callout arearefs="CO1-11">
\r
289 The <literal><min_children></literal> element specifies the minimum number of child
\r
290 processes that should be running at any given time.
\r
293 <callout arearefs="CO1-12">
\r
295 The <literal><max_children></literal> element specifies the maximum number of child
\r
296 processes that should be running at any given time.
\r
299 <callout arearefs="CO1-13">
\r
301 The <literal><min_spare_children></literal> element specifies the minimum number of idle
\r
302 child processes that should be available to handle incoming requests. If there
\r
303 are fewer than this number of spare child processes, new processes will be
\r
307 <callout arearefs="CO1-14">
\r
309 The`<max_spare_children>` element specifies the maximum number of idle
\r
310 child processes that should be available to handle incoming requests. If there
\r
311 are more than this number of spare child processes, the extra processes will be
\r
316 <simpara>To make the service accessible via the public router, you must also
\r
317 edit the <literal>opensrf_core.xml</literal> configuration file to add the service to the list
\r
318 of publicly accessible services:</simpara>
\r
319 <formalpara><title>Making a service publicly accessible in <literal>opensrf_core.xml</literal></title><para>
\r
320 <programlisting language="xml" linenumbering="unnumbered">
\r
321 <router><co id="CO2-1"/>
\r
322 <!-- This is the public router. On this router, we only register applications
\r
323 which should be accessible to everyone on the opensrf network -->
\r
324 <name>router</name>
\r
325 <domain>public.localhost</domain><co id="CO2-2"/>
\r
327 <service>opensrf.math</service>
\r
328 <service>opensrf.simple-text</service> <co id="CO2-3"/>
\r
332 </para></formalpara>
\r
334 <callout arearefs="CO2-1">
\r
336 This section of the <literal>opensrf_core.xml</literal> file is located at XPath
\r
337 <literal>/config/opensrf/routers/</literal>.
\r
340 <callout arearefs="CO2-2">
\r
342 <literal>public.localhost</literal> is the canonical public router domain in the OpenSRF
\r
343 installation instructions.
\r
346 <callout arearefs="CO2-3">
\r
348 Each <literal><service></literal> element contained in the <literal><services></literal> element
\r
349 offers their services via the public router as well as the private router.
\r
353 <simpara>Once you have defined the new service, you must restart the OpenSRF Router
\r
354 to retrieve the new configuration and start or restart the service itself.</simpara>
\r
356 <simplesect id="_calling_an_opensrf_method">
\r
357 <title>Calling an OpenSRF method</title>
\r
358 <simpara>OpenSRF clients in any supported language can invoke OpenSRF services in any
\r
359 supported language. So let’s see a few examples of how we can call our fancy
\r
360 new <literal>opensrf.simple-text.reverse()</literal> method:</simpara>
\r
361 <simplesect id="_calling_opensrf_methods_from_the_srfsh_client">
\r
362 <title>Calling OpenSRF methods from the srfsh client</title>
\r
363 <simpara><literal>srfsh</literal> is a command-line tool installed with OpenSRF that you can use to call
\r
364 OpenSRF methods. To call an OpenSRF method, issue the <literal>request</literal> command and
\r
365 pass the OpenSRF service and method name as the first two arguments; then pass
\r
366 one or more JSON objects delimited by commas as the arguments to the method
\r
367 being invoked.</simpara>
\r
368 <simpara>The following example calls the <literal>opensrf.simple-text.reverse</literal> method of the
\r
369 <literal>opensrf.simple-text</literal> OpenSRF service, passing the string <literal>"foobar"</literal> as the
\r
370 only method argument:</simpara>
\r
371 <programlisting language="sh" linenumbering="unnumbered">
\r
373 srfsh # request opensrf.simple-text opensrf.simple-text.reverse "foobar"
\r
375 Received Data: "raboof"
\r
377 =------------------------------------
\r
378 Request Completed Successfully
\r
379 Request Time in seconds: 0.016718
\r
380 =------------------------------------
\r
383 <simplesect id="opensrfIntrospection">
\r
384 <title>Getting documentation for OpenSRF methods from the srfsh client</title>
\r
385 <simpara>The <literal>srfsh</literal> client also gives you command-line access to retrieving metadata
\r
386 about OpenSRF services and methods. For a given OpenSRF method, for example,
\r
387 you can retrieve information such as the minimum number of required arguments,
\r
388 the data type and a description of each argument, the package or library in
\r
389 which the method is implemented, and a description of the method. To retrieve
\r
390 the documentation for an opensrf method from <literal>srfsh</literal>, issue the <literal>introspect</literal>
\r
391 command, followed by the name of the OpenSRF service and (optionally) the
\r
392 name of the OpenSRF method. If you do not pass a method name to the <literal>introspect</literal>
\r
393 command, <literal>srfsh</literal> lists all of the methods offered by the service. If you pass
\r
394 a partial method name, <literal>srfsh</literal> lists all of the methods that match that portion
\r
395 of the method name.</simpara>
\r
396 <note><simpara>The quality and availability of the descriptive information for each
\r
397 method depends on the developer to register the method with complete and
\r
398 accurate information. The quality varies across the set of OpenSRF and
\r
399 Evergreen APIs, although some effort is being put towards improving the
\r
400 state of the internal documentation.</simpara></note>
\r
401 <programlisting language="sh" linenumbering="unnumbered">
\r
402 srfsh# introspect opensrf.simple-text "opensrf.simple-text.reverse"
\r
403 --> opensrf.simple-text
\r
406 "__c":"opensrf.simple-text",
\r
409 "stream":0, <co id="CO3-1"/>
\r
410 "object_hint":"OpenSRF_Application_Demo_SimpleText",
\r
412 "package":"OpenSRF::Application::Demo::SimpleText", <co id="CO3-2"/>
\r
413 "api_name":"opensrf.simple-text.reverse",<co id="CO3-3"/>
\r
414 "server_class":"opensrf.simple-text",
\r
415 "signature":{ <co id="CO3-4"/>
\r
416 "params":[ <co id="CO3-5"/>
\r
418 "desc":"The string to reverse",
\r
423 "desc":"Returns the input string in reverse order\n", <co id="CO3-6"/>
\r
424 "return":{ <co id="CO3-7"/>
\r
425 "desc":"Returns the input string in reverse order",
\r
429 "method":"text_reverse", <co id="CO3-8"/>
\r
430 "argc":1 <co id="CO3-9"/>
\r
435 <callout arearefs="CO3-1">
\r
437 <literal>stream</literal> denotes whether the method supports streaming responses or not.
\r
440 <callout arearefs="CO3-2">
\r
442 <literal>package</literal> identifies which package or library implements the method.
\r
445 <callout arearefs="CO3-3">
\r
447 <literal>api_name</literal> identifies the name of the OpenSRF method.
\r
450 <callout arearefs="CO3-4">
\r
452 <literal>signature</literal> is a hash that describes the parameters for the method.
\r
455 <callout arearefs="CO3-5">
\r
457 <literal>params</literal> is an array of hashes describing each parameter in the method;
\r
458 each parameter has a description (<literal>desc</literal>), name (<literal>name</literal>), and type (<literal>type</literal>).
\r
461 <callout arearefs="CO3-6">
\r
463 <literal>desc</literal> is a string that describes the method itself.
\r
466 <callout arearefs="CO3-7">
\r
468 <literal>return</literal> is a hash that describes the return value for the method; it
\r
469 contains a description of the return value (<literal>desc</literal>) and the type of the
\r
470 returned value (<literal>type</literal>).
\r
473 <callout arearefs="CO3-8">
\r
475 <literal>method</literal> identifies the name of the function or method in the source
\r
479 <callout arearefs="CO3-9">
\r
481 <literal>argc</literal> is an integer describing the minimum number of arguments that
\r
482 must be passed to this method.
\r
487 <simplesect id="_calling_opensrf_methods_from_perl_applications">
\r
488 <title>Calling OpenSRF methods from Perl applications</title>
\r
489 <simpara>To call an OpenSRF method from Perl, you must connect to the OpenSRF service,
\r
490 issue the request to the method, and then retrieve the results.</simpara>
\r
491 <programlisting language="perl" linenumbering="unnumbered">
\r
494 use OpenSRF::AppSession;
\r
495 use OpenSRF::System;
\r
497 OpenSRF::System->bootstrap_client(config_file => '/openils/conf/opensrf_core.xml');<co id="CO4-1"/>
\r
499 my $session = OpenSRF::AppSession->create("opensrf.simple-text");<co id="CO4-2"/>
\r
501 print "substring: Accepts a string and a number as input, returns a string\n";
\r
502 my $result = $session->request("opensrf.simple-text.substring", "foobar", 3);<co id="CO4-3"/>
\r
503 my $request = $result->gather(); <co id="CO4-4"/>
\r
504 print "Substring: $request\n\n";
\r
506 print "split: Accepts two strings as input, returns an array of strings\n";
\r
507 $request = $session->request("opensrf.simple-text.split", "This is a test", " ");<co id="CO4-5"/>
\r
508 my $output = "Split: [";
\r
510 while ($element = $request->recv()) { <co id="CO4-6"/>
\r
511 $output .= $element->content . ", "; <co id="CO4-7"/>
\r
513 $output =~ s/, $/]/;
\r
514 print $output . "\n\n";
\r
516 print "statistics: Accepts an array of strings as input, returns a hash\n";
\r
517 my @many_strings = [
\r
518 "First I think I'll have breakfast",
\r
519 "Then I think that lunch would be nice",
\r
520 "And then seventy desserts to finish off the day"
\r
523 $result = $session->request("opensrf.simple-text.statistics", @many_strings); <co id="CO4-8"/>
\r
524 $request = $result->gather(); <co id="CO4-9"/>
\r
525 print "Length: " . $result->{'length'} . "\n";
\r
526 print "Word count: " . $result->{'word_count'} . "\n";
\r
528 $session->disconnect(); <co id="CO4-10"/>
\r
531 <callout arearefs="CO4-1">
\r
533 The <literal>OpenSRF::System->bootstrap_client()</literal> method reads the OpenSRF
\r
534 configuration information from the indicated file and creates an XMPP client
\r
535 connection based on that information.
\r
538 <callout arearefs="CO4-2">
\r
540 The <literal>OpenSRF::AppSession->create()</literal> method accepts one argument - the name
\r
541 of the OpenSRF service to which you want to want to make one or more requests -
\r
542 and returns an object prepared to use the client connection to make those
\r
546 <callout arearefs="CO4-3">
\r
548 The <literal>OpenSRF::AppSession->request()</literal> method accepts a minimum of one
\r
549 argument - the name of the OpenSRF method to which you want to make a request -
\r
550 followed by zero or more arguments to pass to the OpenSRF method as input
\r
551 values. This example passes a string and an integer to the
\r
552 <literal>opensrf.simple-text.substring</literal> method defined by the <literal>opensrf.simple-text</literal>
\r
556 <callout arearefs="CO4-4">
\r
558 The <literal>gather()</literal> method, called on the result object returned by the
\r
559 <literal>request()</literal> method, iterates over all of the possible results from the result
\r
560 object and returns a single variable.
\r
563 <callout arearefs="CO4-5">
\r
565 This <literal>request()</literal> call passes two strings to the <literal>opensrf.simple-text.split</literal>
\r
566 method defined by the <literal>opensrf.simple-text</literal> OpenSRF service and returns (via
\r
567 <literal>gather()</literal>) a reference to an array of results.
\r
570 <callout arearefs="CO4-6">
\r
572 The <literal>opensrf.simple-text.split()</literal> method is a streaming method that
\r
573 returns an array of results with one element per <literal>recv()</literal> call on the
\r
574 result object. We could use the <literal>gather()</literal> method to retrieve all of the
\r
575 results in a single array reference, but instead we simply iterate over
\r
576 the result variable until there are no more results to retrieve.
\r
579 <callout arearefs="CO4-7">
\r
581 While the <literal>gather()</literal> convenience method returns only the content of the
\r
582 complete set of results for a given request, the <literal>recv()</literal> method returns an
\r
583 OpenSRF result object with <literal>status</literal>, <literal>statusCode</literal>, and <literal>content</literal> fields as
\r
584 we saw in <link linkend="OpenSRFOverHTTP">the HTTP results example</link>.
\r
587 <callout arearefs="CO4-8">
\r
589 This <literal>request()</literal> call passes an array to the
\r
590 <literal>opensrf.simple-text.statistics</literal> method defined by the <literal>opensrf.simple-text</literal>
\r
594 <callout arearefs="CO4-9">
\r
596 The result object returns a hash reference via <literal>gather()</literal>. The hash
\r
597 contains the <literal>length</literal> and <literal>word_count</literal> keys we defined in the method.
\r
600 <callout arearefs="CO4-10">
\r
602 The <literal>OpenSRF::AppSession->disconnect()</literal> method closes the XMPP client
\r
603 connection and cleans up resources associated with the session.
\r
609 <simplesect id="_accepting_and_returning_more_interesting_data_types">
\r
610 <title>Accepting and returning more interesting data types</title>
\r
611 <simpara>Of course, the example of accepting a single string and returning a single
\r
612 string is not very interesting. In real life, our applications tend to pass
\r
613 around multiple arguments, including arrays and hashes. Fortunately, OpenSRF
\r
614 makes that easy to deal with; in Perl, for example, returning a reference to
\r
615 the data type does the right thing. In the following example of a method that
\r
616 returns a list, we accept two arguments of type string: the string to be split,
\r
617 and the delimiter that should be used to split the string.</simpara>
\r
618 <formalpara><title>Basic text splitting method</title><para>
\r
619 <programlisting language="perl" linenumbering="unnumbered">
\r
624 my $delimiter = shift || ' ';
\r
626 my @split_text = split $delimiter, $text;
\r
627 return \@split_text;
\r
630 __PACKAGE__->register_method(
\r
631 method => 'text_split',
\r
632 api_name => 'opensrf.simple-text.split'
\r
635 </para></formalpara>
\r
636 <simpara>We simply return a reference to the list, and OpenSRF does the rest of the work
\r
637 for us to convert the data into the language-independent format that is then
\r
638 returned to the caller. As a caller of a given method, you must rely on the
\r
639 documentation used to register to determine the data structures - if the developer has
\r
640 added the appropriate documentation.</simpara>
\r
642 <simplesect id="_accepting_and_returning_evergreen_objects">
\r
643 <title>Accepting and returning Evergreen objects</title>
\r
644 <simpara>OpenSRF is agnostic about objects; its role is to pass JSON back and forth
\r
645 between OpenSRF clients and services, and it allows the specific clients and
\r
646 services to define their own semantics for the JSON structures. On top of that
\r
647 infrastructure, Evergreen offers the fieldmapper: an object-relational mapper
\r
648 that provides a complete definition of all objects, their properties, their
\r
649 relationships to other objects, the permissions required to create, read,
\r
650 update, or delete objects of that type, and the database table or view on which
\r
651 they are based.</simpara>
\r
652 <simpara>The Evergreen fieldmapper offers a great deal of convenience for working with
\r
653 complex system objects beyond the basic mapping of classes to database
\r
654 schemas. Although the result is passed over the wire as a JSON object
\r
655 containing the indicated fields, fieldmapper-aware clients then turn those
\r
656 JSON objects into native objects with setter / getter methods for each field.</simpara>
\r
657 <simpara>All of this metadata about Evergreen objects is defined in the
\r
658 fieldmapper configuration file (<literal>/openils/conf/fm_IDL.xml</literal>), and access to
\r
659 these classes is provided by the <literal>open-ils.cstore</literal>, <literal>open-ils.pcrud</literal>, and
\r
660 <literal>open-ils.reporter-store</literal> OpenSRF services which parse the fieldmapper
\r
661 configuration file and dynamically register OpenSRF methods for creating,
\r
662 reading, updating, and deleting all of the defined classes.</simpara>
\r
663 <formalpara><title>Example fieldmapper class definition for "Open User Summary"</title><para>
\r
664 <programlisting language="xml" linenumbering="unnumbered">
\r
665 <class id="mous" controller="open-ils.cstore open-ils.pcrud"
\r
666 oils_obj:fieldmapper="money::open_user_summary"
\r
667 oils_persist:tablename="money.open_usr_summary"
\r
668 reporter:label="Open User Summary"> <co id="CO5-1"/>
\r
669 <fields oils_persist:primary="usr" oils_persist:sequence=""> <co id="CO5-2"/>
\r
670 <field name="balance_owed" reporter:datatype="money" /> <co id="CO5-3"/>
\r
671 <field name="total_owed" reporter:datatype="money" />
\r
672 <field name="total_paid" reporter:datatype="money" />
\r
673 <field name="usr" reporter:datatype="link"/>
\r
676 <link field="usr" reltype="has_a" key="id" map="" class="au"/><co id="CO5-4"/>
\r
678 <permacrud xmlns="http://open-ils.org/spec/opensrf/IDL/permacrud/v1"><co id="CO5-5"/>
\r
680 <retrieve permission="VIEW_USER"><co id="CO5-6"/>
\r
681 <context link="usr" field="home_ou"/><co id="CO5-7"/>
\r
687 </para></formalpara>
\r
689 <callout arearefs="CO5-1">
\r
691 The <literal><class></literal> element defines the class:
\r
696 The <literal>id</literal> attribute defines the <emphasis>class hint</emphasis> that identifies the class both
\r
697 elsewhere in the fieldmapper configuration file, such as in the value of the
\r
698 <literal>field</literal> attribute of the <literal><link></literal> element, and in the JSON object itself when
\r
699 it is instantiated. For example, an "Open User Summary" JSON object would have
\r
700 the top level property of <literal>"__c":"mous"</literal>.
\r
705 The <literal>controller</literal> attribute identifies the services that have direct access
\r
706 to this class. If <literal>open-ils.pcrud</literal> is not listed, for example, then there is
\r
707 no means to directly access members of this class through a public service.
\r
712 The <literal>oils_obj:fieldmapper</literal> attribute defines the name of the Perl
\r
713 fieldmapper class that will be dynamically generated to provide setter and
\r
714 getter methods for instances of the class.
\r
719 The <literal>oils_persist:tablename</literal> attribute identifies the schema name and table
\r
720 name of the database table that stores the data that represents the instances
\r
721 of this class. In this case, the schema is <literal>money</literal> and the table is
\r
722 <literal>open_usr_summary</literal>.
\r
727 The <literal>reporter:label</literal> attribute defines a human-readable name for the class
\r
728 used in the reporting interface to identify the class. These names are defined
\r
729 in English in the fieldmapper configuration file; however, they are extracted
\r
730 so that they can be translated and served in the user’s language of choice.
\r
735 <callout arearefs="CO5-2">
\r
737 The <literal><fields></literal> element lists all of the fields that belong to the object.
\r
742 The <literal>oils_persist:primary</literal> attribute identifies the field that acts as the
\r
743 primary key for the object; in this case, the field with the name <literal>usr</literal>.
\r
748 The <literal>oils_persist:sequence</literal> attribute identifies the sequence object
\r
749 (if any) in this database provides values for new instances of this class. In
\r
750 this case, the primary key is defined by a field that is linked to a different
\r
751 table, so no sequence is used to populate these instances.
\r
756 <callout arearefs="CO5-3">
\r
758 Each <literal><field></literal> element defines a single field with the following attributes:
\r
763 The <literal>name</literal> attribute identifies the column name of the field in the
\r
764 underlying database table as well as providing a name for the setter / getter
\r
765 method that can be invoked in the JSON or native version of the object.
\r
770 The <literal>reporter:datatype</literal> attribute defines how the reporter should treat
\r
771 the contents of the field for the purposes of querying and display.
\r
776 The <literal>reporter:label</literal> attribute can be used to provide a human-readable name
\r
777 for each field; without it, the reporter falls back to the value of the <literal>name</literal>
\r
783 <callout arearefs="CO5-4">
\r
785 The <literal><links></literal> element contains a set of zero or more <literal><link></literal> elements,
\r
786 each of which defines a relationship between the class being described and
\r
792 The <literal>field</literal> attribute identifies the field named in this class that links
\r
793 to the external class.
\r
798 The <literal>reltype</literal> attribute identifies the kind of relationship between the
\r
799 classes; in the case of <literal>has_a</literal>, each value in the <literal>usr</literal> field is guaranteed
\r
800 to have a corresponding value in the external class.
\r
805 The <literal>key</literal> attribute identifies the name of the field in the external
\r
806 class to which this field links.
\r
811 The rarely-used <literal>map</literal> attribute identifies a second class to which
\r
812 the external class links; it enables this field to define a direct
\r
813 relationship to an external class with one degree of separation, to
\r
814 avoid having to retrieve all of the linked members of an intermediate
\r
815 class just to retrieve the instances from the actual desired target class.
\r
820 The <literal>class</literal> attribute identifies the external class to which this field
\r
826 <callout arearefs="CO5-5">
\r
828 The <literal><permacrud></literal> element defines the permissions that must have been
\r
829 granted to a user to operate on instances of this class.
\r
832 <callout arearefs="CO5-6">
\r
834 The <literal><retrieve></literal> element is one of four possible children of the
\r
835 <literal><actions></literal> element that define the permissions required for each action:
\r
836 create, retrieve, update, and delete.
\r
841 The <literal>permission</literal> attribute identifies the name of the permission that must
\r
842 have been granted to the user to perform the action.
\r
847 The <literal>contextfield</literal> attribute, if it exists, defines the field in this class
\r
848 that identifies the library within the system for which the user must have
\r
849 prvileges to work. If a user has been granted a given permission, but has not been
\r
850 granted privileges to work at a given library, they can not perform the action
\r
856 <callout arearefs="CO5-7">
\r
858 The rarely-used <literal><context></literal> element identifies a linked field (<literal>link</literal>
\r
859 attribute) in this class which links to an external class that holds the field
\r
860 (<literal>field</literal> attribute) that identifies the library within the system for which the
\r
861 user must have privileges to work.
\r
865 <simpara>When you retrieve an instance of a class, you can ask for the result to
\r
866 <emphasis>flesh</emphasis> some or all of the linked fields of that class, so that the linked
\r
867 instances are returned embedded directly in your requested instance. In that
\r
868 same request you can ask for the fleshed instances to in turn have their linked
\r
869 fields fleshed. By bundling all of this into a single request and result
\r
870 sequence, you can avoid the network overhead of requiring the client to request
\r
871 the base object, then request each linked object in turn.</simpara>
\r
872 <simpara>You can also iterate over a collection of instances and set the automatically
\r
873 generated <literal>isdeleted</literal>, <literal>isupdated</literal>, or <literal>isnew</literal> properties to indicate that
\r
874 the given instance has been deleted, updated, or created respectively.
\r
875 Evergreen can then act in batch mode over the collection to perform the
\r
876 requested actions on any of the instances that have been flagged for action.</simpara>
\r
878 <simplesect id="_returning_streaming_results">
\r
879 <title>Returning streaming results</title>
\r
880 <simpara>In the previous implementation of the <literal>opensrf.simple-text.split</literal> method, we
\r
881 returned a reference to the complete array of results. For small values being
\r
882 delivered over the network, this is perfectly acceptable, but for large sets of
\r
883 values this can pose a number of problems for the requesting client. Consider a
\r
884 service that returns a set of bibliographic records in response to a query like
\r
885 "all records edited in the past month"; if the underlying database is
\r
886 relatively active, that could result in thousands of records being returned as
\r
887 a single network request. The client would be forced to block until all of the
\r
888 results are returned, likely resulting in a significant delay, and depending on
\r
889 the implementation, correspondingly large amounts of memory might be consumed
\r
890 as all of the results are read from the network in a single block.</simpara>
\r
891 <simpara>OpenSRF offers a solution to this problem. If the method returns results that
\r
892 can be divided into separate meaningful units, you can register the OpenSRF
\r
893 method as a streaming method and enable the client to loop over the results one
\r
894 unit at a time until the method returns no further results. In addition to
\r
895 registering the method with the provided name, OpenSRF also registers an additional
\r
896 method with <literal>.atomic</literal> appended to the method name. The <literal>.atomic</literal> variant gathers
\r
897 all of the results into a single block to return to the client, giving the caller
\r
898 the ability to choose either streaming or atomic results from a single method
\r
899 definition.</simpara>
\r
900 <simpara>In the following example, the text splitting method has been reimplemented to
\r
901 support streaming; very few changes are required:</simpara>
\r
902 <formalpara><title>Text splitting method - streaming mode</title><para>
\r
903 <programlisting language="perl" linenumbering="unnumbered">
\r
908 my $delimiter = shift || ' ';
\r
910 my @split_text = split $delimiter, $text;
\r
911 foreach my $string (@split_text) { <co id="CO6-1"/>
\r
912 $conn->respond($string);
\r
917 __PACKAGE__->register_method(
\r
918 method => 'text_split',
\r
919 api_name => 'opensrf.simple-text.split',
\r
920 stream => 1<co id="CO6-2"/>
\r
923 </para></formalpara>
\r
925 <callout arearefs="CO6-1">
\r
927 Rather than returning a reference to the array, a streaming method loops
\r
928 over the contents of the array and invokes the <literal>respond()</literal> method of the
\r
929 connection object on each element of the array.
\r
932 <callout arearefs="CO6-2">
\r
934 Registering the method as a streaming method instructs OpenSRF to also
\r
935 register an atomic variant (<literal>opensrf.simple-text.split.atomic</literal>).
\r
940 <simplesect id="_error_warning_info_debug">
\r
941 <title>Error! Warning! Info! Debug!</title>
\r
942 <simpara>As hard as it may be to believe, it is true: applications sometimes do not
\r
943 behave in the expected manner, particularly when they are still under
\r
944 development. The service language bindings for OpenSRF include integrated
\r
945 support for logging messages at the levels of ERROR, WARNING, INFO, DEBUG, and
\r
946 the extremely verbose INTERNAL to either a local file or to a syslogger
\r
947 service. The destination of the log files, and the level of verbosity to be
\r
948 logged, is set in the <literal>opensrf_core.xml</literal> configuration file. To add logging to
\r
949 our Perl example, we just have to add the <literal>OpenSRF::Utils::Logger</literal> package to our
\r
950 list of used Perl modules, then invoke the logger at the desired logging level.</simpara>
\r
951 <simpara>You can include many calls to the OpenSRF logger; only those that are higher
\r
952 than your configured logging level will actually hit the log. The following
\r
953 example exercises all of the available logging levels in OpenSRF:</simpara>
\r
954 <programlisting language="perl" linenumbering="unnumbered">
\r
955 use OpenSRF::Utils::Logger;
\r
956 my $logger = OpenSRF::Utils::Logger;
\r
957 # some code in some function
\r
959 $logger->error("Hmm, something bad DEFINITELY happened!");
\r
960 $logger->warn("Hmm, something bad might have happened.");
\r
961 $logger->info("Something happened.");
\r
962 $logger->debug("Something happened; here are some more details.");
\r
963 $logger->internal("Something happened; here are all the gory details.")
\r
966 <simpara>If you call the mythical OpenSRF method containing the preceding OpenSRF logger
\r
967 statements on a system running at the default logging level of INFO, you will
\r
968 only see the INFO, WARN, and ERR messages, as follows:</simpara>
\r
969 <formalpara><title>Results of logging calls at the default level of INFO</title><para>
\r
971 [2010-03-17 22:27:30] opensrf.simple-text [ERR :5681:SimpleText.pm:277:] Hmm, something bad DEFINITELY happened!
\r
972 [2010-03-17 22:27:30] opensrf.simple-text [WARN:5681:SimpleText.pm:278:] Hmm, something bad might have happened.
\r
973 [2010-03-17 22:27:30] opensrf.simple-text [INFO:5681:SimpleText.pm:279:] Something happened.
\r
975 </para></formalpara>
\r
976 <simpara>If you then increase the the logging level to INTERNAL (5), the logs will
\r
977 contain much more information, as follows:</simpara>
\r
978 <formalpara><title>Results of logging calls at the default level of INTERNAL</title><para>
\r
980 [2010-03-17 22:48:11] opensrf.simple-text [ERR :5934:SimpleText.pm:277:] Hmm, something bad DEFINITELY happened!
\r
981 [2010-03-17 22:48:11] opensrf.simple-text [WARN:5934:SimpleText.pm:278:] Hmm, something bad might have happened.
\r
982 [2010-03-17 22:48:11] opensrf.simple-text [INFO:5934:SimpleText.pm:279:] Something happened.
\r
983 [2010-03-17 22:48:11] opensrf.simple-text [DEBG:5934:SimpleText.pm:280:] Something happened; here are some more details.
\r
984 [2010-03-17 22:48:11] opensrf.simple-text [INTL:5934:SimpleText.pm:281:] Something happened; here are all the gory details.
\r
985 [2010-03-17 22:48:11] opensrf.simple-text [ERR :5934:SimpleText.pm:283:] Resolver did not find a cache hit
\r
986 [2010-03-17 22:48:21] opensrf.simple-text [INTL:5934:Cache.pm:125:] Stored opensrf.simple-text.test_cache.masaa => "here" in memcached server
\r
987 [2010-03-17 22:48:21] opensrf.simple-text [DEBG:5934:Application.pm:579:] Coderef for [OpenSRF::Application::Demo::SimpleText::test_cache]...
\r
988 [2010-03-17 22:48:21] opensrf.simple-text [DEBG:5934:Application.pm:586:] A top level Request object is responding de nada
\r
989 [2010-03-17 22:48:21] opensrf.simple-text [DEBG:5934:Application.pm:190:] Method duration for [opensrf.simple-text.test_cache]: 10.005
\r
990 [2010-03-17 22:48:21] opensrf.simple-text [INTL:5934:AppSession.pm:780:] Calling queue_wait(0)
\r
991 [2010-03-17 22:48:21] opensrf.simple-text [INTL:5934:AppSession.pm:769:] Resending...0
\r
992 [2010-03-17 22:48:21] opensrf.simple-text [INTL:5934:AppSession.pm:450:] In send
\r
993 [2010-03-17 22:48:21] opensrf.simple-text [DEBG:5934:AppSession.pm:506:] AppSession sending RESULT to opensrf@private.localhost/...
\r
994 [2010-03-17 22:48:21] opensrf.simple-text [DEBG:5934:AppSession.pm:506:] AppSession sending STATUS to opensrf@private.localhost/...
\r
997 </para></formalpara>
\r
998 <simpara>To see everything that is happening in OpenSRF, try leaving your logging level
\r
999 set to INTERNAL for a few minutes - just ensure that you have a lot of free disk
\r
1000 space available if you have a moderately busy system!</simpara>
\r
1002 <simplesect id="_caching_results_one_secret_of_scalability">
\r
1003 <title>Caching results: one secret of scalability</title>
\r
1004 <simpara>If you have ever used an application that depends on a remote Web service
\r
1005 outside of your control — say, if you need to retrieve results from a
\r
1006 microblogging service — you know the pain of latency and dependability (or the
\r
1007 lack thereof). To improve the response time for OpenSRF services, you can take
\r
1008 advantage of the support offered by the <literal>OpenSRF::Utils::Cache</literal> module for
\r
1009 communicating with a local instance or cluster of <literal>memcache</literal> daemons to store
\r
1010 and retrieve persistent values. The following example demonstrates caching
\r
1011 by sleeping for 10 seconds the first time it receives a given cache key and
\r
1012 cannot retrieve a corresponding value from the cache:</simpara>
\r
1013 <formalpara><title>Simple caching OpenSRF service</title><para>
\r
1014 <programlisting language="perl" linenumbering="unnumbered">
\r
1015 use OpenSRF::Utils::Cache;<co id="CO7-1"/>
\r
1019 my $test_key = shift;
\r
1020 my $cache = OpenSRF::Utils::Cache->new('global'); <co id="CO7-2"/>
\r
1021 my $cache_key = "opensrf.simple-text.test_cache.$test_key"; <co id="CO7-3"/>
\r
1022 my $result = $cache->get_cache($cache_key) || undef; <co id="CO7-4"/>
\r
1024 $logger->info("Resolver found a cache hit");
\r
1027 sleep 10; <co id="CO7-5"/>
\r
1028 my $cache_timeout = 300; <co id="CO7-6"/>
\r
1029 $cache->put_cache($cache_key, "here", $cache_timeout); <co id="CO7-7"/>
\r
1030 return "There was no cache hit.";
\r
1033 </para></formalpara>
\r
1035 <callout arearefs="CO7-1">
\r
1037 The OpenSRF::Utils::Cache module provides access to the built-in caching
\r
1038 support in OpenSRF.
\r
1041 <callout arearefs="CO7-2">
\r
1043 The constructor for the cache object accepts a single argument to define
\r
1044 the cache type for the object. Each cache type can use a separate <literal>memcache</literal>
\r
1045 server to keep the caches separated. Most Evergreen services use the <literal>global</literal>
\r
1046 cache, while the <literal>anon</literal> cache is used for Web sessions.
\r
1049 <callout arearefs="CO7-3">
\r
1051 The cache key is simply a string that uniquely identifies the value you
\r
1052 want to store or retrieve. This line creates a cache key based on the OpenSRF
\r
1053 method name and request input value.
\r
1056 <callout arearefs="CO7-4">
\r
1058 The <literal>get_cache()</literal> method checks to see if the cache key already exists. If
\r
1059 a matching key is found, the service immediately returns the stored value.
\r
1062 <callout arearefs="CO7-5">
\r
1064 If the cache key does not exist, the code sleeps for 10 seconds to
\r
1065 simulate a call to a slow remote Web service or an intensive process.
\r
1068 <callout arearefs="CO7-6">
\r
1070 The <literal>$cache_timeout</literal> variable represents a value for the lifetime of the
\r
1071 cache key in seconds.
\r
1074 <callout arearefs="CO7-7">
\r
1076 After the code retrieves its value (or, in the case of this example,
\r
1077 finishes sleeping), it creates the cache entry by calling the <literal>put_cache()</literal>
\r
1078 method. The method accepts three arguments: the cache key, the value to be
\r
1079 stored ("here"), and the timeout value in seconds to ensure that we do not
\r
1080 return stale data on subsequent calls.
\r
1085 <simplesect id="_initializing_the_service_and_its_children_child_labour">
\r
1086 <title>Initializing the service and its children: child labour</title>
\r
1087 <simpara>When an OpenSRF service is started, it looks for a procedure called
\r
1088 <literal>initialize()</literal> to set up any global variables shared by all of the children of
\r
1089 the service. The <literal>initialize()</literal> procedure is typically used to retrieve
\r
1090 configuration settings from the <literal>opensrf.xml</literal> file.</simpara>
\r
1091 <simpara>An OpenSRF service spawns one or more children to actually do the work
\r
1092 requested by callers of the service. For every child process an OpenSRF service
\r
1093 spawns, the child process clones the parent environment and then each child
\r
1094 process runs the <literal>child_init()</literal> process (if any) defined in the OpenSRF service
\r
1095 to initialize any child-specific settings.</simpara>
\r
1096 <simpara>When the OpenSRF service kills a child process, it invokes the <literal>child_exit()</literal>
\r
1097 procedure (if any) to clean up any resources associated with the child process.
\r
1098 Similarly, when the OpenSRF service is stopped, it calls the <literal>DESTROY()</literal>
\r
1099 procedure to clean up any remaining resources.</simpara>
\r
1101 <simplesect id="_retrieving_configuration_settings">
\r
1102 <title>Retrieving configuration settings</title>
\r
1103 <simpara>The settings for OpenSRF services are maintained in the <literal>opensrf.xml</literal> XML
\r
1104 configuration file. The structure of the XML document consists of a root
\r
1105 element <literal><opensrf></literal> containing two child elements:</simpara>
\r
1109 The <literal><default></literal> element contains an <literal><apps></literal> element describing all
\r
1110 OpenSRF services running on this system — see <xref linkend="serviceRegistration"/> --, as
\r
1111 well as any other arbitrary XML descriptions required for global configuration
\r
1112 purposes. For example, Evergreen uses this section for email notification and
\r
1113 inter-library patron privacy settings.
\r
1118 The <literal><hosts></literal> element contains one element per host that participates in
\r
1119 this OpenSRF system. Each host element must include an <literal><activeapps></literal> element
\r
1120 that lists all of the services to start on this host when the system starts
\r
1121 up. Each host element can optionally override any of the default settings.
\r
1125 <simpara>OpenSRF includes a service named <literal>opensrf.settings</literal> to provide distributed
\r
1126 cached access to the configuration settings with a simple API:</simpara>
\r
1130 <literal>opensrf.settings.default_config.get</literal> accepts zero arguments and returns
\r
1131 the complete set of default settings as a JSON document.
\r
1136 <literal>opensrf.settings.host_config.get</literal> accepts one argument (hostname) and
\r
1137 returns the complete set of settings, as customized for that hostname, as a
\r
1143 <literal>opensrf.settings.xpath.get</literal> accepts one argument (an
\r
1144 <ulink url="http://www.w3.org/TR/xpath/">XPath</ulink> expression) and returns the portion of
\r
1145 the configuration file that matches the expression as a JSON document.
\r
1149 <simpara>For example, to determine whether an Evergreen system uses the opt-in
\r
1150 support for sharing patron information between libraries, you could either
\r
1151 invoke the <literal>opensrf.settings.default_config.get</literal> method and parse the
\r
1152 JSON document to determine the value, or invoke the <literal>opensrf.settings.xpath.get</literal>
\r
1153 method with the XPath <literal>/opensrf/default/share/user/opt_in</literal> argument to
\r
1154 retrieve the value directly.</simpara>
\r
1155 <simpara>In practice, OpenSRF includes convenience libraries in all of its client
\r
1156 language bindings to simplify access to configuration values. C offers
\r
1157 osrfConfig.c, Perl offers <literal>OpenSRF::Utils::SettingsClient</literal>, Java offers
\r
1158 <literal>org.opensrf.util.SettingsClient</literal>, and Python offers <literal>osrf.set</literal>. These
\r
1159 libraries locally cache the configuration file to avoid network roundtrips for
\r
1160 every request and enable the developer to request specific values without
\r
1161 having to manually construct XPath expressions.</simpara>
\r
1164 <section id="_getting_under_the_covers_with_opensrf">
\r
1165 <title>OpenSRF Communication Flows</title>
\r
1166 <simpara>Now that you have seen that it truly is easy to create an OpenSRF service, we
\r
1167 can take a look at what is going on under the covers to make all of this work
\r
1168 for you.</simpara>
\r
1169 <simplesect id="_get_on_the_messaging_bus_safely">
\r
1170 <title>Get on the messaging bus - safely</title>
\r
1171 <simpara>One of the core innovations of OpenSRF was to use the Extensible Messaging and
\r
1172 Presence Protocol (XMPP, more colloquially known as Jabber) as the messaging
\r
1173 bus that ties OpenSRF services together across servers. XMPP is an "XML
\r
1174 protocol for near-real-time messaging, presence, and request-response services"
\r
1175 (<ulink url="http://www.ietf.org/rfc/rfc3920.txt">http://www.ietf.org/rfc/rfc3920.txt</ulink>) that OpenSRF relies on to handle most of
\r
1176 the complexity of networked communications. OpenSRF requres an XMPP server
\r
1177 that supports multiple domains such as <ulink url="http://www.ejabberd.im/">ejabberd</ulink>.
\r
1178 Multiple domain support means that a single server can support XMPP virtual
\r
1179 hosts with separate sets of users and access privileges per domain. By
\r
1180 routing communications through separate public and private XMPP domains,
\r
1181 OpenSRF services gain an additional layer of security.</simpara>
\r
1182 <simpara>The <ulink url="http://evergreen-ils.org/dokuwiki/doku.php?id=opensrf:1.2:install">OpenSRF
\r
1183 installation documentation</ulink> instructs you to create two separate hostnames
\r
1184 (<literal>private.localhost</literal> and <literal>public.localhost</literal>) to use as XMPP domains. OpenSRF
\r
1185 can control access to its services based on the domain of the client and
\r
1186 whether a given service allows access from clients on the public domain. When
\r
1187 you start OpenSRF, the first XMPP clients that connect to the XMPP server are
\r
1188 the OpenSRF public and private <emphasis>routers</emphasis>. OpenSRF routers maintain a list of
\r
1189 available services and connect clients to available services. When an OpenSRF
\r
1190 service starts, it establishes a connection to the XMPP server and registers
\r
1191 itself with the private router. The OpenSRF configuration contains a list of
\r
1192 public OpenSRF services, each of which must also register with the public
\r
1195 <simplesect id="_opensrf_communication_flows_over_xmpp">
\r
1196 <title>OpenSRF communication flows over XMPP</title>
\r
1197 <simpara>In a minimal OpenSRF deployment, two XMPP users named "router" connect to the
\r
1198 XMPP server, with one connected to the private XMPP domain and one connected to
\r
1199 the public XMPP domain. Similarly, two XMPP users named "opensrf" connect to
\r
1200 the XMPP server via the private and public XMPP domains. When an OpenSRF
\r
1201 service is started, it uses the "opensrf" XMPP user to advertise its
\r
1202 availability with the corresponding router on that XMPP domain; the XMPP server
\r
1203 automatically assigns a Jabber ID (<emphasis>JID</emphasis>) based on the client hostname to each
\r
1204 service’s listener process and each connected drone process waiting to carry
\r
1205 out requests. When an OpenSRF router receives a request to invoke a method on a
\r
1206 given service, it connects the requester to the next available listener in the
\r
1207 list of registered listeners for that service.</simpara>
\r
1208 <simpara>Services and clients connect to the XMPP server using a single set of XMPP
\r
1209 client credentials (for example, <literal>opensrf@private.localhost</literal>), but use XMPP
\r
1210 resource identifiers to differentiate themselves in the JID for each
\r
1211 connection. For example, the JID for a copy of the <literal>opensrf.simple-text</literal>
\r
1212 service with process ID <literal>6285</literal> that has connected to the <literal>private.localhost</literal>
\r
1213 domain using the <literal>opensrf</literal> XMPP client credentials could be
\r
1214 <literal>opensrf@private.localhost/opensrf.simple-text_drone_at_localhost_6285</literal>. By
\r
1215 convention, the user name for OpenSRF clients is <literal>opensrf</literal>, and the user name
\r
1216 for OpenSRF routers is <literal>router</literal>, so the XMPP server for OpenSRF will have four
\r
1217 separate users registered:
\r
1218 * <literal>opensrf@private.localhost</literal> is an OpenSRF client that connects with these
\r
1219 credentials and which can access any OpenSRF service.
\r
1220 * <literal>opensrf@public.localhost</literal> is an OpenSRF client that connects with these
\r
1221 credentials and which can only access OpenSRF services that have registered
\r
1222 with the public router.
\r
1223 * <literal>router@private.localhost</literal> is the private OpenSRF router with which all
\r
1224 services register.
\r
1225 * <literal>router@public.localhost</literal> is the public OpenSRF router with which only
\r
1226 services that must be publicly accessible register.</simpara>
\r
1227 <simpara>All OpenSRF services automatically register themselves with the private XMPP
\r
1228 domain, but only those services that register themselves with the public XMPP
\r
1229 domain can be invoked from public OpenSRF clients. The OpenSRF client and
\r
1230 router user names, passwords, and domain names, along with the list of services
\r
1231 that should be public, are contained in the <literal>opensrf_core.xml</literal> configuration
\r
1234 <simplesect id="OpenSRFOverHTTP">
\r
1235 <title>OpenSRF communication flows over HTTP</title>
\r
1236 <simpara>In some contexts, access to a full XMPP client is not a practical option. For
\r
1237 example, while XMPP clients have been implemented in JavaScript, you might
\r
1238 be concerned about browser compatibility and processing overhead - or you might
\r
1239 want to issue OpenSRF requests from the command line with <literal>curl</literal>. Fortunately,
\r
1240 any OpenSRF service registered with the public router is accessible via the
\r
1241 OpenSRF HTTP Translator. The OpenSRF HTTP Translator implements the
\r
1242 <ulink url="http://www.open-ils.org/dokuwiki/doku.php?id=opensrf_over_http">OpenSRF-over-HTTP
\r
1243 proposed specification</ulink> as an Apache module that translates HTTP requests into
\r
1244 OpenSRF requests and returns OpenSRF results as HTTP results to the initiating
\r
1245 HTTP client.</simpara>
\r
1246 <formalpara><title>Issuing an HTTP POST request to an OpenSRF method via the OpenSRF HTTP Translator</title><para>
\r
1247 <programlisting language="bash" linenumbering="unnumbered">
\r
1248 # curl request broken up over multiple lines for legibility
\r
1249 curl -H "X-OpenSRF-service: opensrf.simple-text"<co id="CO8-1"/>
\r
1250 --data 'osrf-msg=[ \<co id="CO8-2"/>
\r
1251 {"__c":"osrfMessage","__p":{"threadTrace":0,"locale":"en-CA", <co id="CO8-3"/>
\r
1252 "type":"REQUEST","payload": {"__c":"osrfMethod","__p":
\r
1253 {"method":"opensrf.simple-text.reverse","params":["foobar"]}
\r
1256 http://localhost/osrf-http-translator <co id="CO8-4"/>
\r
1258 </para></formalpara>
\r
1260 <callout arearefs="CO8-1">
\r
1262 The <literal>X-OpenSRF-service</literal> header identifies the OpenSRF service of interest.
\r
1265 <callout arearefs="CO8-2">
\r
1267 The POST request consists of a single parameter, the <literal>osrf-msg</literal> value,
\r
1268 which contains a JSON array.
\r
1271 <callout arearefs="CO8-3">
\r
1273 The first object is an OpenSRF message (<literal>"__c":"osrfMessage"</literal>) with a set of
\r
1274 parameters (<literal>"__p":{}</literal>).
\r
1279 The identifier for the request (<literal>"threadTrace":0</literal>); this value is echoed
\r
1280 back in the result.
\r
1285 The message type (<literal>"type":"REQUEST"</literal>).
\r
1290 The locale for the message; if the OpenSRF method is locale-sensitive, it
\r
1291 can check the locale for each OpenSRF request and return different information
\r
1292 depending on the locale.
\r
1297 The payload of the message (<literal>"payload":{}</literal>) containing the OpenSRF method
\r
1298 request (<literal>"__c":"osrfMethod"</literal>) and its parameters (<literal>"__p:"{}</literal>).
\r
1303 The method name for the request (<literal>"method":"opensrf.simple-text.reverse"</literal>).
\r
1308 A set of JSON parameters to pass to the method (<literal>"params":["foobar"]</literal>); in
\r
1309 this case, a single string <literal>"foobar"</literal>.
\r
1316 <callout arearefs="CO8-4">
\r
1318 The URL on which the OpenSRF HTTP translator is listening,
\r
1319 <literal>/osrf-http-translator</literal> is the default location in the Apache example
\r
1320 configuration files shipped with the OpenSRF source, but this is configurable.
\r
1324 <formalpara><title>Results from an HTTP POST request to an OpenSRF method via the OpenSRF HTTP Translator</title><para>
\r
1325 <programlisting language="bash" linenumbering="unnumbered">
\r
1326 # HTTP response broken up over multiple lines for legibility
\r
1327 [{"__c":"osrfMessage","__p": <co id="CO9-1"/>
\r
1328 {"threadTrace":0, "payload": <co id="CO9-2"/>
\r
1329 {"__c":"osrfResult","__p": <co id="CO9-3"/>
\r
1330 {"status":"OK","content":"raboof","statusCode":200} <co id="CO9-4"/>
\r
1331 },"type":"RESULT","locale":"en-CA" <co id="CO9-5"/>
\r
1334 {"__c":"osrfMessage","__p": <co id="CO9-6"/>
\r
1335 {"threadTrace":0,"payload": <co id="CO9-7"/>
\r
1336 {"__c":"osrfConnectStatus","__p": <co id="CO9-8"/>
\r
1337 {"status":"Request Complete","statusCode":205}<co id="CO9-9"/>
\r
1338 },"type":"STATUS","locale":"en-CA" <co id="CO9-10"/>
\r
1342 </para></formalpara>
\r
1344 <callout arearefs="CO9-1">
\r
1346 The OpenSRF HTTP Translator returns an array of JSON objects in its
\r
1347 response. Each object in the response is an OpenSRF message
\r
1348 (<literal>"__c":"osrfMessage"</literal>) with a collection of response parameters (<literal>"__p":</literal>).
\r
1351 <callout arearefs="CO9-2">
\r
1353 The OpenSRF message identifier (<literal>"threadTrace":0</literal>) confirms that this
\r
1354 message is in response to the request matching the same identifier.
\r
1357 <callout arearefs="CO9-3">
\r
1359 The message includes a payload JSON object (<literal>"payload":</literal>) with an OpenSRF
\r
1360 result for the request (<literal>"__c":"osrfResult"</literal>).
\r
1363 <callout arearefs="CO9-4">
\r
1365 The result includes a status indicator string (<literal>"status":"OK"</literal>), the content
\r
1366 of the result response - in this case, a single string "raboof"
\r
1367 (<literal>"content":"raboof"</literal>) - and an integer status code for the request
\r
1368 (<literal>"statusCode":200</literal>).
\r
1371 <callout arearefs="CO9-5">
\r
1373 The message also includes the message type (<literal>"type":"RESULT"</literal>) and the
\r
1374 message locale (<literal>"locale":"en-CA"</literal>).
\r
1377 <callout arearefs="CO9-6">
\r
1379 The second message in the set of results from the response.
\r
1382 <callout arearefs="CO9-7">
\r
1384 Again, the message identifier confirms that this message is in response to
\r
1385 a particular request.
\r
1388 <callout arearefs="CO9-8">
\r
1390 The payload of the message denotes that this message is an
\r
1391 OpenSRF connection status message (<literal>"__c":"osrfConnectStatus"</literal>), with some
\r
1392 information about the particular OpenSRF connection that was used for this
\r
1396 <callout arearefs="CO9-9">
\r
1398 The response parameters for an OpenSRF connection status message include a
\r
1399 verbose status (<literal>"status":"Request Complete"</literal>) and an integer status code for
\r
1400 the connection status (`"statusCode":205).
\r
1403 <callout arearefs="CO9-10">
\r
1405 The message also includes the message type (<literal>"type":"RESULT"</literal>) and the
\r
1406 message locale (<literal>"locale":"en-CA"</literal>).
\r
1410 <tip><simpara>Before adding a new public OpenSRF service, ensure that it does
\r
1411 not introduce privilege escalation or unchecked access to data. For example,
\r
1412 the Evergreen <literal>open-ils.cstore</literal> private service is an object-relational mapper
\r
1413 that provides read and write access to the entire Evergreen database, so it
\r
1414 would be catastrophic to expose that service publicly. In comparison, the
\r
1415 Evergreen <literal>open-ils.pcrud</literal> public service offers the same functionality as
\r
1416 <literal>open-ils.cstore</literal> to any connected HTTP client or OpenSRF client, but the
\r
1417 additional authentication and authorization layer in <literal>open-ils.pcrud</literal> prevents
\r
1418 unchecked access to Evergreen’s data.</simpara></tip>
\r
1420 <simplesect id="_stateless_and_stateful_connections">
\r
1421 <title>Stateless and stateful connections</title>
\r
1422 <simpara>OpenSRF supports both <emphasis>stateless</emphasis> and <emphasis>stateful</emphasis> connections. When an OpenSRF
\r
1423 client issues a <literal>REQUEST</literal> message in a <emphasis>stateless</emphasis> connection, the router
\r
1424 forwards the request to the next available service and the service returns the
\r
1425 result directly to the client.</simpara>
\r
1427 <simpara>When an OpenSRF client issues a <literal>CONNECT</literal> message to create a <emphasis>stateful</emphasis> conection, the
\r
1428 router returns the Jabber ID of the next available service to the client so
\r
1429 that the client can issue one or more <literal>REQUEST</literal> message directly to that
\r
1430 particular service and the service will return corresponding <literal>RESULT</literal> messages
\r
1431 directly to the client. Until the client issues a <literal>DISCONNECT</literal> message, that
\r
1432 particular service is only available to the requesting client. Stateful connections
\r
1433 are useful for clients that need to make many requests from a particular service,
\r
1434 as it avoids the intermediary step of contacting the router for each request, as
\r
1435 well as for operations that require a controlled sequence of commands, such as a
\r
1436 set of database INSERT, UPDATE, and DELETE statements within a transaction.</simpara>
\r
1439 <simplesect id="_message_body_format">
\r
1440 <title>Message body format</title>
\r
1441 <simpara>OpenSRF was an early adopter of JavaScript Object Notation (JSON). While XMPP
\r
1442 is an XML protocol, the Evergreen developers recognized that the compactness of
\r
1443 the JSON format offered a significant reduction in bandwidth for the volume of
\r
1444 messages that would be generated in an application of that size. In addition,
\r
1445 the ability of languages such as JavaScript, Perl, and Python to generate
\r
1446 native objects with minimal parsing offered an attractive advantage over
\r
1447 invoking an XML parser for every message. Instead, the body of the XMPP message
\r
1448 is a simple JSON structure. For a simple request, like the following example
\r
1449 that simply reverses a string, it looks like a significant overhead: but we get
\r
1450 the advantages of locale support and tracing the request from the requester
\r
1451 through the listener and responder (drone).</simpara>
\r
1452 <formalpara><title>A request for opensrf.simple-text.reverse("foobar"):</title><para>
\r
1453 <programlisting language="xml" linenumbering="unnumbered">
\r
1454 <message from='router@private.localhost/opensrf.simple-text'
\r
1455 to='opensrf@private.localhost/opensrf.simple-text_listener_at_localhost_6275'
\r
1456 router_from='opensrf@private.localhost/_karmic_126678.3719_6288'
\r
1457 router_to='' router_class='' router_command='' osrf_xid=''
\r
1459 <thread>1266781414.366573.12667814146288</thread>
\r
1462 {"__c":"osrfMessage","__p":
\r
1463 {"threadTrace":"1","locale":"en-US","type":"REQUEST","payload":
\r
1464 {"__c":"osrfMethod","__p":
\r
1465 {"method":"opensrf.simple-text.reverse","params":["foobar"]}
\r
1473 </para></formalpara>
\r
1474 <formalpara><title>A response from opensrf.simple-text.reverse("foobar")</title><para>
\r
1475 <programlisting language="xml" linenumbering="unnumbered">
\r
1476 <message from='opensrf@private.localhost/opensrf.simple-text_drone_at_localhost_6285'
\r
1477 to='opensrf@private.localhost/_karmic_126678.3719_6288'
\r
1478 router_command='' router_class='' osrf_xid=''
\r
1480 <thread>1266781414.366573.12667814146288</thread>
\r
1483 {"__c":"osrfMessage","__p":
\r
1484 {"threadTrace":"1","payload":
\r
1485 {"__c":"osrfResult","__p":
\r
1486 {"status":"OK","content":"raboof","statusCode":200}
\r
1487 } ,"type":"RESULT","locale":"en-US"}
\r
1489 {"__c":"osrfMessage","__p":
\r
1490 {"threadTrace":"1","payload":
\r
1491 {"__c":"osrfConnectStatus","__p":
\r
1492 {"status":"Request Complete","statusCode":205}
\r
1493 },"type":"STATUS","locale":"en-US"}
\r
1499 </para></formalpara>
\r
1500 <simpara>The content of the <literal><body></literal> element of the OpenSRF request and result should
\r
1501 look familiar; they match the structure of the <link linkend="OpenSRFOverHTTP">OpenSRF over HTTP examples</link> that we previously dissected.</simpara>
\r
1503 <simplesect id="_registering_opensrf_methods_in_depth">
\r
1504 <title>Registering OpenSRF methods in depth</title>
\r
1505 <simpara>Let’s explore the call to <literal>__PACKAGE__->register_method()</literal>; most of the members
\r
1506 of the hash are optional, and for the sake of brevity we omitted them in the
\r
1507 previous example. As we have seen in the results of the <link linkend="opensrfIntrospection">introspection call</link>, a
\r
1508 verbose registration method call is recommended to better enable the internal
\r
1509 documentation. Here is the complete set of members that you should pass to
\r
1510 <literal>__PACKAGE__->register_method()</literal>:</simpara>
\r
1514 The <literal>method</literal> member specifies the name of the procedure in this module that is being registered as an OpenSRF method.
\r
1519 The <literal>api_name</literal> member specifies the invocable name of the OpenSRF method; by convention, the OpenSRF service name is used as the prefix.
\r
1524 The optional <literal>api_level</literal> member can be used for versioning the methods to allow the use of a deprecated API, but in practical use is always 1.
\r
1529 The optional <literal>argc</literal> member specifies the minimal number of arguments that the method expects.
\r
1534 The optional <literal>stream</literal> member, if set to any value, specifies that the method supports returning multiple values from a single call to
\r
1535 subsequent requests. OpenSRF automatically creates a corresponding method with ".atomic" appended to its name that returns the complete set of results in a
\r
1536 single request. Streaming methods are useful if you are returning hundreds of records and want to act on the results as they return.
\r
1541 The optional <literal>signature</literal> member is a hash that describes the method’s purpose, arguments, and return value.
\r
1546 The <literal>desc</literal> member of the <literal>signature</literal> hash describes the method’s purpose.
\r
1551 The <literal>params</literal> member of the <literal>signature</literal> hash is an array of hashes in which each array element describes the corresponding method
\r
1552 argument in order.
\r
1557 The <literal>name</literal> member of the argument hash specifies the name of the argument.
\r
1562 The <literal>desc</literal> member of the argument hash describes the argument’s purpose.
\r
1567 The <literal>type</literal> member of the argument hash specifies the data type of the argument: for example, string, integer, boolean, number, array, or hash.
\r
1574 The <literal>return</literal> member of the <literal>signature</literal> hash is a hash that describes the return value of the method.
\r
1579 The <literal>desc</literal> member of the <literal>return</literal> hash describes the return value.
\r
1584 The <literal>type</literal> member of the <literal>return</literal> hash specifies the data type of the return value: for example, string, integer, boolean, number,
\r
1595 <section id="_evergreen_specific_opensrf_services">
\r
1596 <title>Evergreen-specific OpenSRF services</title>
\r
1597 <simpara>Evergreen is currently the primary showcase for the use of OpenSRF as an
\r
1598 application architecture. Evergreen 1.6.1 includes the following
\r
1599 set of OpenSRF services:</simpara>
\r
1603 The <systemitem class="service">open-ils.actor</systemitem> service supports common tasks for working with user
\r
1604 accounts and libraries.
\r
1609 The <systemitem class="service">open-ils.auth</systemitem> service supports authentication of Evergreen users.
\r
1614 The <systemitem class="service">open-ils.booking</systemitem> service supports the management of reservations
\r
1615 for bookable items.
\r
1620 The <systemitem class="service">open-ils.cat</systemitem> service supports common cataloging tasks, such as
\r
1621 creating, modifying, and merging bibliographic and authority records.
\r
1626 The <systemitem class="service">open-ils.circ</systemitem> service supports circulation tasks such as checking
\r
1627 out items and calculating due dates.
\r
1632 The <systemitem class="service">open-ils.collections</systemitem> service supports tasks that assist collections
\r
1633 agencies in contacting users with outstanding fines above a certain
\r
1639 The <systemitem class="service">open-ils.cstore</systemitem> private service supports unrestricted access to
\r
1640 Evergreen fieldmapper objects.
\r
1645 The <systemitem class="service">open-ils.ingest</systemitem> private service supports tasks for importing
\r
1646 data such as bibliographic and authority records.
\r
1651 The <systemitem class="service">open-ils.pcrud</systemitem> service supports permission-based access to Evergreen
\r
1652 fieldmapper objects.
\r
1657 The <systemitem class="service">open-ils.penalty</systemitem> penalty service supports the calculation of
\r
1658 penalties for users, such as being blocked from further borrowing, for
\r
1659 conditions such as having too many items checked out or too many unpaid
\r
1665 The <systemitem class="service">open-ils.reporter</systemitem> service supports the creation and scheduling of
\r
1671 The <systemitem class="service">open-ils.reporter-store</systemitem> private service supports access to Evergreen
\r
1672 fieldmapper objects for the reporting service.
\r
1677 The <systemitem class="service">open-ils.search</systemitem> service supports searching across bibliographic
\r
1678 records, authority records, serial records, Z39.50 sources, and ZIP codes.
\r
1683 The <systemitem class="service">open-ils.storage</systemitem> private service supports a deprecated method of
\r
1684 providing access to Evergreen fieldmapper objects. Implemented in Perl,
\r
1685 this service has largely been replaced by the much faster C-based
\r
1686 <literal>open-ils.cstore</literal> service.
\r
1691 The <systemitem class="service">open-ils.supercat</systemitem> service supports transforms of MARC records into
\r
1692 other formats, such as MODS, as well as providing Atom and RSS feeds and
\r
1698 The <systemitem class="service">open-ils.trigger</systemitem> private service supports event-based triggers for
\r
1699 actions such as overdue and holds available notification emails.
\r
1704 The <systemitem class="service">open-ils.vandelay</systemitem> service supports the import and export of batches of
\r
1705 bibliographic and authority records.
\r
1709 <simpara>Of some interest is that the <systemitem class="service">open-ils.reporter-store</systemitem> and <systemitem class="service">open-ils.cstore</systemitem>
\r
1710 services have identical implementations. Surfacing them as separate services
\r
1711 enables a deployer of Evergreen to ensure that the reporting service does not
\r
1712 interfere with the performance-critical <systemitem class="service">open-ils.cstore</systemitem> service. One can also
\r
1713 direct the reporting service to a read-only database replica to, again, avoid
\r
1714 interference with <systemitem class="service">open-ils.cstore</systemitem> which must write to the master database.</simpara>
\r
1715 <simpara>There are only a few significant services that are not built on OpenSRF in
\r
1716 Evergreen 1.6.0, such as the SIP and Z39.50 servers. These services implement
\r
1717 different protocols and build on existing daemon architectures (Simple2ZOOM
\r
1718 for Z39.50), but still rely on the other OpenSRF services to provide access
\r
1719 to the Evergreen data. The non-OpenSRF services are reasonably self-contained
\r
1720 and can be deployed on different servers to deliver the same sort of deployment
\r
1721 flexibility as OpenSRF services, but have the disadvantage of not being
\r
1722 integrated into the same configuration and control infrastructure as the
\r
1723 OpenSRF services.</simpara>
\r