After netatalk service has been stopped via "/etc/init.d/atalk stop", the service will not start again by any means other than rebooting the entire system. This also happens when "/etc/init.d/atalk restart" is used. Reproducible: Always Steps to Reproduce: 1. execute "/etc/init.d/atalk stop" 2. execute "/etc/init.d/atalk start" Actual Results: [/etc/init.d/atalk stop] Unregistering morrigan:WorkstationComputer Division...ok Unregistering morrigan:netatalkComputer Division...ok Stopping atalkd...ok Stopping afpd...ok [/etc/init.d/atalk start] Starting atalkd...!! AppleTalk not up! Check your syslog for the reason. Child exited with 1 Registering morrigan:WorkstationComputer Division..!! nbp_rgstr: Connection timed out Can't register morrigan:WorkstationComputer Division@ Registering morrigan:netatalkComputer Division...!! nbp_rgstr: Connection timed out Can't register morrigan:netatalkComputer Division@ Starting afpd..ok Expected Results: [/etc/init.d/atalk stop] Unregistering morrigan:WorkstationComputer Division...ok Unregistering morrigan:netatalkComputer Division...ok Stopping atalkd...ok Stopping afpd...ok [/etc/init.d/atalk start] Starting atalkd... Registering morrigan:WorkstationComputer Division... Registering morrigan:netatalkComputer Division... Starting afpd... Machine: Sun Ultra5 (UltraSPARC IIi 270MHz CPU) 256MB RAM 20GB IDE HD Type 5 Keyboard/Crossbow Mouse Gentoo v2004.0 Looking at netatalk.log, the problem is caused by the stopping of the service not flushing the AppleTalk address. atalkd will re-write the atalkd.conf file in /etc/netatalk on its own. As a test I had the parameters set to "eth0 -phase 2" and then restarted using "/etc/init.d/atalk restart" restarted fine and now "-net 1000-1999 -addr 1769.52 -zone "Computer Division" have been added. If I wanted to restart the service again I would have to go into /etc/netatalk/atald.conf and take out the "-addr 1769.52" and resave the file manually. [Related Log Lines] Apr 09 17:29:42 atalkd[1030][main.c:1090]: I:ATalkDaemon: restart (1.6.4) Apr 09 17:29:43 atalkd[1030][main.c:1366]: E:ATalkDaemon: bind 1278.65:6: Address already in use [emerge info] Gentoo Base System version 1.4.3.13 Portage 2.0.50-r1 (default-sparc64-2004.0, gcc-3.2.3, glibc-2.3.2-r9, 2.4.24- sparc-r2) ================================================================= System uname: 2.4.24-sparc-r2 sparc64 sun4u Autoconf: sys-devel/autoconf-2.58-r1 Automake: sys-devel/automake-1.7.7 ACCEPT_KEYWORDS="sparc" AUTOCLEAN="yes" CFLAGS="-O2 -mcpu=ultrasparc" CHOST="sparc-unknown-linux-gnu" COMPILER="gcc3" CONFIG_PROTECT="/etc /usr/X11R6/lib/X11/xkb /usr/kde/2/share/config /usr/kde/3.2 /share/config /usr/kde/3/share/config /usr/share/config /var/qmail/control" CONFIG_PROTECT_MASK="/etc/gconf /etc/env.d" CXXFLAGS="-O2 -mcpu=ultrasparc" DISTDIR="/usr/portage/distfiles" FEATURES="ccache sandbox" GENTOO_MIRRORS="http://gentoo.oregonstate.edu http://www.ibiblio.org/pub/Linux/distributions/gentoo" MAKEOPTS="-j2" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="X arts avi berkdb crypt cups encode esd fbcon foomaticdb gdbm gif gnome gtk guile imlib java jpeg kde ldap libwww mad mikmod motif mpeg mysql ncurses nls oggvorbis opengl oss pam pdflib perl png postgres python qt readline sdl slang sparc spell ssl tcltk tcpd truetype xml2 xmms xv zlib"
Created attachment 28990 [details] Full netatalk.log file from test restart This log as generated by a single execution of "/etc/init.d/atalk restart"
Created attachment 28991 [details] Copy of atalkd.conf
Any clue on this guys?
Reassigning to net-fs, this isn't really sparc related.
net-fs folks, can you take a look when you get a chance?
I'm not a member of the net-fs group yet, but after having worked on netatalk-2.0.1 ebuilds for a while and having a fair amount of experience with netatalk-2.0.1 in the past, I would like to make some observations. 1) I have never seen the behavior of netatalk trying to change the /etc/netatalk/atalkd.conf file. I trust the reporter that it is, but I wonder what configuration option at either run or compiletime is causing that. That doesn't seem SPARC specific, though, and I would count it as one bug, albeit a non-fatal one without the addition of the other bug. 2) What I would count as the other bug is how the kernel does not seem to be recognizing that atalkd freed its address in the past, hence why it can't start up with the same one as it is trying to. I can attest that I do not see this on IA32 or PowerPC32, so I highly suspect it is a bug in the AppleTalk kernel module on SPARC, or _perhaps_ a netatalk bug that only gets exercised on SPARC, something to do with the way netatalk closes the socket or the way the AppleTalk driver handles a socket being closed. I will look into both of these further, though I am prioritizing the second bug as it is rather the more severe one.
netatalk needs a developer to take up maintenance.
Please try with new netatalk-2.0.3 and report back thanks.
