From citylink.dinoex.sub.org!admin@net2.dinoex.sub.org Mon Jun 9 18:56:32 1997 Received: from mail.Contrib.Com (mail.Contrib.Com [194.77.12.8]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id SAA29538 for ; Mon, 9 Jun 1997 18:56:30 -0700 (PDT) Received: from net2.dinoex.sub.org (net2.dinoex.sub.de [193.203.172.193]) by mail.Contrib.Com (8.8.4/8.8.4) with SMTP id DAA05183 for ; Tue, 10 Jun 1997 03:56:08 +0200 (MET DST) Received: from citylink.dinoex.sub.org by net2.dinoex.sub.org with UUCP (Smail3.1.29.1 #1) id m0wbGE8-000DzoC; Tue, 10 Jun 97 04:00 CEST Received: (from admin@localhost) by citylink.dinoex.sub.org (8.8.5/PMuch-B3b) id DAA17993; Tue, 10 Jun 1997 03:33:45 +0200 (CEST) Message-Id: <199706100133.DAA17993@citylink.dinoex.sub.org> Date: Tue, 10 Jun 1997 03:33:45 +0200 (CEST) From: admin@citylink.dinoex.sub.org (Maschinenwart) Reply-To: admin@citylink.dinoex.sub.org To: FreeBSD-gnats-submit@freebsd.org Subject: Kerberos hangs rcp X-Send-Pr-Version: 3.2 >Number: 3826 >Category: bin >Synopsis: KerberosIV sometimes hangs rcp >Confidential: no >Severity: non-critical >Priority: medium >Responsible: freebsd-bugs >State: closed >Quarter: >Keywords: >Date-Required: >Class: sw-bug >Submitter-Id: current-users >Arrival-Date: Mon Jun 9 19:00:04 PDT 1997 >Closed-Date: Wed Oct 10 12:14:54 PDT 2001 >Last-Modified: Wed Oct 10 12:16:35 PDT 2001 >Originator: Peter Much >Release: FreeBSD 2.2.1-RELEASE i386 >Organization: Beyond the Future Shockwave: Convenience or Conviviality? >Environment: kerberos4 / eBones. >Description: I'm entering an rsh-command that will last some while, f.i.: $ rsh d1 cat /cdrom/2.2.1-RELEASE/ports/ports.tgz | tar xvzf - somestuff During that execution "netstat" on "d1" shows the connection as: tcp 0 0 192.168.98.4.939 192.168.98.11.5119 ESTABLISHED tcp 0 17280 192.168.98.4.544 192.168.98.11.5120 ESTABLISHED Quite after completion I'm entering another rsh-command, f.i.: $ rcp d1:/cdrom/distfiles/something /somewhere rcp: remote host doesn't support Kerberos, using standard rcp Permission denied. The kerberos-support on host d1 seemingly has disappeared now. From now on, any other rcmd-executions from this host to host d1 will deny with the same effect. "netstat -n" on "d1" shows now: tcp 0 0 192.168.98.4.939 192.168.98.11.5119 TIME_WAIT tcp 0 0 192.168.98.4.544 192.168.98.11.5120 TIME_WAIT (On the local host these connections already have disappeared at command-completion.) This will last for ca. 5 minutes, then the TIME_WAIT tcp-connections do disappear and rsh/rcp does work again. >How-To-Repeat: Use rsh-commands that do run for some minutes. If the rsh-commands are short-lived, the connection between port 544 and 5120 gets useable again just after command-completion. >Fix: *** usr.bin/rlogin/kcmd.c-orig Tue Jun 10 00:28:19 1997 --- usr.bin/rlogin/kcmd.c Tue Jun 10 02:53:00 1997 *************** *** 152,157 **** --- 152,158 ---- */ if (errno == ECONNREFUSED && timo <= 4) { /* sleep(timo); don't wait at all here */ + lport--; ! /* timo *= 2; */ continue; } This obviousely will help as a workaround. Somebody with some insight in tcp internals should find out why these connections aren't linked down correctly... regards, PM >Release-Note: >Audit-Trail: Responsible-Changed-From-To: gnats-admin->freebsd-bugs Responsible-Changed-By: max Responsible-Changed-When: Mon Jun 9 23:19:19 PDT 1997 Responsible-Changed-Why: Misfiled PR. State-Changed-From-To: open->closed State-Changed-By: johan State-Changed-When: Wed Oct 10 12:14:54 PDT 2001 State-Changed-Why: The file in the patch was removed 4 years ago. Please open a new PR if the problem persists in more recent versions of FreeBSD. http://www.FreeBSD.org/cgi/query-pr.cgi?pr=3826 >Unformatted: