Update notes for 2019-01-21

content/posts/2019-01.md

@@ -605,4 +605,79 @@ sys     0m2.396s
 - The query currently shows 3023 items, but a [Discovery search for Livestock CRP only returns 858 items](https://cgspace.cgiar.org/discover?filtertype_1=crpsubject&filter_relational_operator_1=equals&filter_1=Livestock&submit_apply_filter=&query=)
 - That query seems to return items tagged with `Livestock and Fish` CRP as well... hmm.
 
+## 2019-01-21
+
+- Investigating running Tomcat 7 on Ubuntu 18.04 with the tarball and a custom systemd package instead of waiting for our DSpace to get compatible with Ubuntu 18.04's Tomcat 8.5
+- I could either run with a simple `tomcat7.service` like this:
+
+```
+[Unit]
+Description=Apache Tomcat 7 Web Application Container
+After=network.target
+[Service]
+Type=forking
+ExecStart=/path/to/apache-tomcat-7.0.92/bin/startup.sh
+ExecStop=/path/to/apache-tomcat-7.0.92/bin/shutdown.sh
+User=aorth
+Group=aorth
+[Install]
+WantedBy=multi-user.target
+```
+
+- Or try to use adapt a real systemd service like Arch Linux's:
+
+```
+[Unit]
+Description=Tomcat 7 servlet container
+After=network.target
+
+[Service]
+Type=forking
+PIDFile=/var/run/tomcat7.pid
+Environment=CATALINA_PID=/var/run/tomcat7.pid
+Environment=TOMCAT_JAVA_HOME=/usr/lib/jvm/default-runtime
+Environment=CATALINA_HOME=/usr/share/tomcat7
+Environment=CATALINA_BASE=/usr/share/tomcat7
+Environment=CATALINA_OPTS=
+Environment=ERRFILE=SYSLOG
+Environment=OUTFILE=SYSLOG
+
+ExecStart=/usr/bin/jsvc \
+            -Dcatalina.home=${CATALINA_HOME} \
+            -Dcatalina.base=${CATALINA_BASE} \
+            -Djava.io.tmpdir=/var/tmp/tomcat7/temp \
+            -cp /usr/share/java/commons-daemon.jar:/usr/share/java/eclipse-ecj.jar:${CATALINA_HOME}/bin/bootstrap.jar:${CATALINA_HOME}/bin/tomcat-juli.jar \
+            -user tomcat7 \
+            -java-home ${TOMCAT_JAVA_HOME} \
+            -pidfile /var/run/tomcat7.pid \
+            -errfile ${ERRFILE} \
+            -outfile ${OUTFILE} \
+            $CATALINA_OPTS \
+            org.apache.catalina.startup.Bootstrap
+
+ExecStop=/usr/bin/jsvc \
+            -pidfile /var/run/tomcat7.pid \
+            -stop \
+            org.apache.catalina.startup.Bootstrap
+
+[Install]
+WantedBy=multi-user.target
+```
+
+- I see that `jsvc` and `libcommons-daemon-java` are both available on Ubuntu so that should be easy to port
+- We probably don't need Eclipse Java Bytecode Compiler (ecj)
+- I tested Tomcat 7.0.92 on Arch Linux using the `tomcat7.service` with `jsvc` and it works... nice!
+- I think I might manage this the same way I do the restic releases in the [Ansible infrastructure scripts](https://github.com/ilri/rmg-ansible-public), where I download a specific version and symlink to some generic location without the version number
+- I verified that there is indeed an issue with sharded Solr statistics cores on DSpace, which will cause inaccurate results in the dspace-statistics-api:
+
+```
+$ http 'http://localhost:3000/solr/statistics/select?indent=on&rows=0&q=type:2+id:11576&fq=isBot:false&fq=statistics_type:view' | grep numFound
+<result name="response" numFound="33" start="0">
+$ http 'http://localhost:3000/solr/statistics-2018/select?indent=on&rows=0&q=type:2+id:11576&fq=isBot:false&fq=statistics_type:view' | grep numFound
+<result name="response" numFound="241" start="0">
+```
+
+- I opened an issue on the GitHub issue tracker ([#10](https://github.com/ilri/dspace-statistics-api/issues/10))
+- I don't think the [SolrClient library](https://solrclient.readthedocs.io/en/latest/) we are currently using supports these type of queries so we might have to just do raw queries with requests
+
 <!-- vim: set sw=2 ts=2: -->