What property or setup parameter determines the URLs displayed on the task attempts webpage of the job/task trackers? My cluster seems to be configured such that all URLs for higher pages (the top cluster admin page, the individual job overview page, and the map/reduce task list page) show URLs by ip address, but the lowest page (the task attempt list for a single task) shows the URLs for the Machine and Task Logs columns by "localhost", not by ip address (although the Counters column still uses the ip address just like URLs on all the higher pages).
The "localhost" links obviously don't work (the cluster is not on the local machine, it's on Tier 3)...unless I just happen to have a cluster also running on my local machine; then the links work but obviously they go to my local machine and thus describe a completely unrelated Hadoop cluster!!! It goes without saying, that's ridiculous. So to get it to work, I have to manually copy/paste the ip address into the URLs every time I want to view those pages...which makes it incredibly tedious to view the task logs. I've asked this a few times now and have gotten no response. Does no one have any idea how to properly configure Hadoop to get around this? I've experimented with the mapred-site.xml mapred.job.tracker and mapred.task.tracker.http.address properties to no avail. What's going on here? <Desperate> ________________________________________________________________________________ Keith Wiley [email protected] keithwiley.com music.keithwiley.com "I used to be with it, but then they changed what it was. Now, what I'm with isn't it, and what's it seems weird and scary to me." -- Abe (Grandpa) Simpson ________________________________________________________________________________
