Red Hat Bugzilla – Bug 39210
linuxconf over xiinetd does not work
Last modified: 2007-04-18 12:33:04 EDT
Description of Problem: linuxconf web interface does not work How Reproducible: always Steps to Reproduce: 1. ntsysv , activate linuxconf 2. linuxconf, activate web access 3. http://localhost:98 Actual Results: endless select, accept loop Expected Results: linuxconf web gui Additional Information:
Same thing happens for me. If I telnet in to localhost port 98 xinetd opens the connection, but offering the HTTP command "GET / HTTP/1.0" doesn't do anything (same happens with Netscape and Lynx). Further, when I kill the telnet session with ^], linuxconf still runs in the background, and according to top, uses all the cpu (98%).
BTW, start linuxconf with $ linuxconf --http then it will listen to port 3000, http://localhost:3000
Ok, that doesn't even come close to working. linuxconf --http does not listen on any port. Running "netstat -l --numeric-ports" even shows that nothing is listening on port 3000. Of course, you can't telnet or point a browser to it either. From my experience with xinetd services (some could be different), they operate on stdin and stdout. Running "linuxconf --http" (which, by the way is what xinetd runs. see /etc/xinetd.d/linuxconf-web) sets it up to receive http commands (thru stdin) and send back html (thru stdout). Neither of these involve listening on any port.
You have to enable the http mode in the linuxconf GUI first!
I am seeing the same looping behavior on RHL 7.1, too. Bugzilla issue #40303 seems relevant, and was helpful to getting the proper configuration established and port assignment done. However, it still doesn't solve the looping problem on my system, although the submitter indicates it did the trick on getting a usable linuxconf working on his setup.
recently posted on red hat updates ftp site xinetd v. 2.3.0 solves the problem of initial browser connection, but when it comes to the Accept change, the browser indicates broken connection (lynx message: "Unexpected network error; connection aborted"). At the same time, xinitd syslog entry pops the message: "warning: can't get client address: Transport endpoint is not connected"
Closing because we don't ship linuxconf anymore
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.