FTS clean and restart procedure

Occassionally, the FTS service scripts can get in strange state. This sometimes happens after a software upgrade where the daemons were left running, but has been noted in other cases.

In this case, the best option is to stop the service, clean the configuration work directories and restart the service.

Follow the procedure below:

1. Stop the service. service gLite stop
2. Delete all watchdog config. rm -i /opt/glite/etc/watchdog.d/*
3. Delete all logrotate config. rm -i /etc/logrotate.d/*glite*
4. kill -9 all processes owend by user sc3 that you find (ps aux | grep sc3 to find them)
5. kill -9 all processes owned by user tomcat4 that you find
6. Delete the tomcat gLite config. rm -i /etc/tomcat5/Catalina/localhost/*glite*
7. Delete the tomcat work directory. rm -rf /usr/share/tomcat5/work/*
8. Delete the tomcat webapp cache for glite applications. cd /usr/share/tomcat5/webapps/. Delete rm -rf all directories except webdav, balancer and ROOT.
9. Delete the intermediate agent config. rm -i /opt/glite/etc/glite-transfer-agent-fts-*.properties.xml /opt/glite/etc/glite-transfer-agent-fts-*.log-properties

this will clean the environment. Then rerun the configuration script:

/opt/glite/etc/config/scripts/glite-file-transfer-service-oracle-config.py

which will rebuild the sub-config files, and attempt to restart the service.

TODO: this should be scripted. Getting a * wrong could result in nasty problems.

-- GavinMcCance - 07 Jul 2005