source: vlo/trunk/DEPLOY-README.txt @ 6102

Last change on this file since 6102 was 6102, checked in by Twan Goosen, 9 years ago

extended deployment and upgrade instructions

File size: 7.3 KB
Line 
1The instructions below describe the installation of the applications necessary
2for browsing: the VLO web application, the Solr server, and the meta data
3importer.
4
51. The archive
6
7   The VLO importer and web application are contained in an
8   archive. In the case of version 3.0, the archive is named
9
10        vlo-3.0-Distribution.tar.gz
11
12   Deploying the VLO application means:
13   - unpacking the archive in a suitable location
14   - unpacking the war files contained in the archive
15   - adapt the application's context files
16   and
17   - adapt the VloConfig.xml main configuration file.
18   - optionally executing upgrade steps described in the UPGRADE file
19
202. Archive unpacking
21
22   Unpack the archive, for example
23
24         vlo-3.0-Distribution.tar.gz
25
26   in a temporary directory: 'temp'. Next, stop the Tomcat server:
27
28        /etc/init.d/tomcat6 stop
29
30   After the server has stopped, copy the contents of the vlo_parent-2.13
31   directory just created to a permanent directory, 'vlo', for example. In the
32   tree starting in vlo, the configuration of the application is stored. Also,
33   from the
34
35        vlo/bin
36   
37   directory, you can run the importer application. Since this application
38   relies on the Solr server, we first need to install the Solr web application.
39
40   Recursively assign ownership of the entire unpacked distribution to the appropriate
41   user (i.e. 'vlouser' on catalog.clarin.eu).
42
433. Solr server installation
44
45   The VLO importer and web application use Solr as their database
46   server. To install this server, assuming that the Tomcat server
47   has not started yet, copy the Solr server web application archive
48   to the Tomcat server web application directory. This could for
49   example be
50
51        /var/lib/tomcat6/webapps
52
53   Step into this directory, lets call it 'apps',
54
55        cd apps
56        cp temp/vlo-3.0/war/vlo-solr-3.0.war .
57
58   If it does not exist, create the vlo_solr directory, and unpack the web
59   application archive in it:
60
61        cd vlo_solr
62        unzip ../vlo-solr-3.0.war 
63
64   After unzipping, remove the vlo-solr-3.0.war file. Next, if necessary,
65   modify the solr/home parameter in the
66 
67        apps/META-INF/context.xml
68
69   file to the path where the solr server finds its configuration and stores
70   its data:
71
72        vlo/config/solr
73
74   Solr needs a location to store its index data. This needs to be configured
75   through a Java system property 'solr.data.dir'. Configure this in your
76   Tomcat instance, for example by adding the following to
77   ${catalina.home}/bin/setenv.sh:
78   
79    export JAVA_OPTS="$JAVA_OPTS -Dsolr.data.dir=/var/vlo/solr/data"
80
81   The directory does not have to exist, but its parent does and the Tomcat
82   user needs write access in that location.
83   
84   Copy the new context.xml file to the Tomcat configuration:
85
86        cp apps/META-INF/context.xml \
87                $CATALINA_HOME/conf/Catalina/localhost/vlo_solr.xml
88
89   If you would like logging to be configured different from the type of
90   logging packaged, please modify
91
92        apps/WEB-INF/classes/log4j.properties
93   
94   If necessary, change the ownership of the files in the tree starting in
95
96        vlo/config/solr
97
98   to that of the Tomcat user. This will enable the Solr server can store data
99   in it. Now the Solr server has been installed, the importer and web
100   application could use it. Next we will install the VLO web application.
101
1024. Web application installation
103
104   Similar to the Solr archive, unpack
105
106        temp/vlo-3.0/war/vlo-web-app-3.0.war in
107   
108   the 
109
110        apps/vlo
111
112   directory.
113
114   Because the packaged configuration is suitable to very specific (development)
115   circumstances only, in most cases it does fit the production environment. You
116   can specify an alternative, external configuration file, preferably the same
117   file as the one used by the importer (see 5). Please modify
118
119        apps/META-INF/context.xml
120
121   by adding a reference to an external configuration file. By modifying this
122   file, you can adapt the VLO configuration to your needs. You might, for
123   example, need to assign another value to the solrUrl parameter also. This
124   parameter is used to let the VLO web application know where it can reach the
125   Solr server.
126
127   Instead of changing the value of the solrUrl parameter in the external
128   configuration file, you could also supply an alternative value by adding
129   the parameter to
130
131        apps/META-INF/context.xml
132
133   directly. The comments in this file will tell you how to add the parameter.
134
135   Like in the case of the Solr server, copy the apps/META-INF/context.xml
136   context file to:
137
138        /var/lib/tomcat6/conf/Catalina/localhost/vlo.xml
139
140   or to another, comparable path.
141
142   If you like, you can change the web application's way of logging. This time,
143   modifications should be applied to:
144
145        apps/WEB-INF/classes/log4j.properties
146
147   Now the web application has been installed and configured, we only need to
148   have a look at the VLO importer's configuration.
149
1505. Importer configuration
151
152   As mentioned in the description of the web application installation, a
153   typical setup needs an external configuration file. For consistency reasons
154   both the web application and application that imports data into the Solr
155   data base should use one and the same configuration file. So if, for example,
156   the web application's configuration is in
157
158        $VLO_CONFIG/VloConfig.xml, use
159
160        ./vlo_solr_importer.sh $VLO_CONFIG/VloConfig.xml
161
162   to run the importer in 'bin'. A default configuration file is supplied as
163   'config/VloConfig.xml'. The importer script will default to this location
164   if no location is specified on the command line.
165
166   Most likely, in the configuration file, the dataRoot values need to be
167   changed. This is what a dataRoot definition could look like:
168
169           <originName>MPI self harvest</originName>
170                 <rootFile>/var/www/vlomd/self/</rootFile>           
171                 <prefix>http://m12404423/vlomd/</prefix>
172                 <tostrip>/var/www/vlomd/</tostrip>
173                 <deleteFirst>false</deleteFirst>
174           </DataRoot>
175
176   A dataRoot element describes the meta data files. The toStrip part of the
177   description is left out of the rootFile part to create a http link to the
178   meta data; the links starts with the prefix.
179
180   Apart from the dataRoot values, the solrUrl parameter might need to be
181   changed. Please note that the context path value defined in
182
183        $CATALINA_HOME/conf/Catalina/localhost/vlo_solr.xml
184
185   or in a file equivalent to this one, should be reflected in the value of
186   this parameter.
187
188   Whenever the definition of the database changes, it is recommended to clear
189   the database before starting an importer run. To do this, make sure that the
190   'deletaAllFirst' parameter equals true before the run. After that, the
191   original value of the parameter could be restored.
192
1936. Importing data
194
195   Before starting data import, first start the Tomcat server:
196
197        /etc/init.d/tomcat6 start
198
199   To be sure the Solr server is working as it should, inspect the Tomcat log
200   file directory, and in that directory, check the solr.log and vlo.log files.
201
202   Next, you can run the importer by starting the
203
204        ./vlo_solr_importer.sh
205
206   script in the vlo/bin directory. Messages are logged to the console. Because
207   meta data is not static, it is recommended to run the importer a couple of
208   times a week. Please note that, given the current (04.2014) set of data, a 
209   run typically takes between two and four hours.
Note: See TracBrowser for help on using the repository browser.