Feel free to open a ticket. But this cannot be done at the script level - there's a race condition if you were to do that. So any fix needs to be in the AgentRun class.
Karl On Tue, Jul 5, 2011 at 12:13 AM, Shinichiro Abe <[email protected]> wrote: > Thank you. > I set up sync directory property, then it worked well. > > BTW, though I don't know whether it has relevant to the synchronization, > can it check running more than one of instance of agent? > For example, if one runs "executecommand.sh ~ agents.AgentRun" twice, > can the command warn that it is already running? > Now, if one runs the command twice, Java procceses run twice. > > Thank you, > Shinichiro Abe > > On 2011/07/05, at 9:26, Karl Wright wrote: > >> It sounds like you have not set up synchronization properly for the >> multi-process installation you are running. See: >> http://incubator.apache.org/connectors/how-to-build-and-deploy.html#Examples >> >> The synch directory needs to be specified for multi-process installations. >> >> Karl >> >> >> On Mon, Jul 4, 2011 at 9:37 AM, Shinichiro Abe >> <[email protected]> wrote: >>> Hi. >>> I scheduled the job of once crawling.I set including *.txt only on Paths >>> tab. >>> I started the job, text files were indexed. >>> After that, I changed *.txt into *.xls. >>> I started the job, it should index xls files, but text files were not >>> deleted and xls files were not indexed. >>> >>> This problem occurs in Tomcat + agent process, not in Jetty. >>> (If restarting agent process after changing target file, this can be >>> resolved.) >>> Which is correct behavior? >>> It seems that it is raised by cache managing, or looks like some bugs >>> around seeding >>> or something wrong on my Tomcat environment. >>> >>> And it occurs on both once crawling and continuous crawling. >>> And it occurs on both jcifs and filesystem. >>> >>> Regards, Shichiro Abe > >
