[daisy] Notes from updating daisy from 1.4.1 to 1.5

[Hack Kampbjørn]

Last week I updated our daisy installation from 1.4.1 to 1.5. It mostly 
went without problems here are my notes about the update.

* ActiveMQ and InnoDB: the template configuration file enforces InnoDB 
engine in MySQL sessionVariables=storage_engine=InnoDB, but this is 
disabled on our MySQL server. I couldn't find any note for this in Daisy or 
ActiveMQ documentation. My guess is because of transactions. Should I move 
this database to our PostgreSQL server instead?

* file permissions: I only found one implicit reference to file permissions 
(using p when extracting the tarball).

I ended with this setting:
chown -R daisy:daisy ${DAISY_HOME}
chmod -R u=rX,g=rX,o=rX ${DAISY_HOME}
chown -R daisy:operator ${DAISY_DATADIR}
chmod -R u=rwX,g=rwX,o= ${DAISY_DATADIR}
chown -R tomcat:operator ${DAISY_WIKIDATA}
chmod -R u=rX,g=rwX,o= ${DAISY_WIKIDATA}
chmod -R u=rwX,g=rX,o= ${DAISY_WIKIDATA}/logs
chmod -R u=rwX,g=rwX,o= ${DAISY_WIKIDATA}/bookstore
chmod -R u=rwX,g=rwX,o= ${DAISY_WIKIDATA}/books

NOTE: that DAISY_HOME/repository-server/merlin does also need write 
permission when using the standard scripts but not when using tanuki 
wrapper.

* daisywiki.data and log files. The log configuration WEB-INF/log4j.xconf 
and WEB-INF/logkit.xconf did not pickup daisywiki.data from application 
context but logkit.xconf did work when using a system property. log4j.xconf 
doesn't seem to be used at least there is nothing logged to the log4j file 
yet. (We don't use the embedded Jetty but run the web application in Tomcat)

* tanuki wrapper: I had several problems with the wrapper configuration, 
all easy to fix things. In the wrapper script:
- change all tail -1 to tail -n 1t. Using tail -1 on gentoo linux gives 
deprecated warning messages
- change PIDDIR from . to ${DAISY_DATADIR}. daisy may not have write 
permission to . but does to DAISY_DATADIR
- change RUN_AS_GROUP=RUN_AS_USER to RUN_AS_GROUP=$RUN_AS_USER
- change su -m to just su.

The wrapper did not seem to do environment expansion on the configuration 
file so I run it through sed first:
sed \
        -e "s#%JAVA_HOME%#${JAVA_HOME}#" \
        -e "s#%DAISY_HOME%#${DAISY_HOME}#" \
        -e "s#%DAISY_DATADIR%#${DAISY_DATADIR}#" \
        -e "s#%WRAPPER_HOME%#${WRAPPER_HOME}#" \
< ${WRAPPER_HOME}/conf/daisy-repository.conf.in > 
${WRAPPER_HOME}/conf/daisy-repository.conf


This is not an update problem but I tried to upload a 50Mb PDF document, 
first I needed to increase the upload limit for the Wiki 
<http://cocoondev.org/daisydocs-1_5/admin/217.html>, but it was not 
automatically indexed. I found this in the log:

[INFO   ] <2006-09-21 16:23:34,984> 
(daisy.repository.fulltextindexupdater): Will not extract text from 
document ID 238,
 branch ID 1, language ID 2, part ID 4 because the part data is too large 
(49719624 > 20000000)

/Hack