Package Details: netatalk 3.1.13-3

Git Clone URL: (read-only, click to copy)
Package Base: netatalk
Description: Open-source implementation of the Apple Filing Protocol
Upstream URL:
Keywords: afp appletalk
Licenses: GPL
Replaces: netatalk-git, netatalk2
Submitter: Snowman
Maintainer: denn
Last Packager: denn
Votes: 95
Popularity: 0.000211
First Submitted: 2007-03-31 06:19 (UTC)
Last Updated: 2022-05-29 13:34 (UTC)

Latest Comments

jac_cbi commented on 2022-05-31 15:25 (UTC)

I can confirm that the latest update to the AUR (3.1.13-3) does indeed work. This is the relevant fix:

@@ -1637,14 +1650,21 @@ void *ad_entry(const struct adouble *ad, int eid)
        size_t len = ad_getentrylen(ad, eid);
        bool valid;

+       if (bufsize == 0) {
+               bufsize = sizeof(ad->ad_data) - (off + len);
+       }
        valid = ad_entry_check_size(eid, bufsize, off, len);
        if (!valid) {

Which looks like it does fix the problem, but I'm really hoping upstream digs deeper and locates the initialization of ad where ad->valid_data_len is left at zero or the call site where it's changed to zero.

jac_cbi commented on 2022-05-27 15:09 (UTC)

With the patch more complete patch from FreeBSD (just more logging info, no other logic changes from first patch), I get the following logs:

May 27 10:58:49.152479 afpd[18141] {ad_open.c:1639} (error:ad): ad_entry_check_size overflow (122 + 32 > 0)
May 27 10:58:49.152605 afpd[18141] {ad_open.c:1655} (error:ad): ad_entry: not valid

So, it would appear that ad->valid_data_len is 0 prior to ad_entry_check_size() getting called. I don't know enough about the code to say wether that's normal or not. :-/

Anywho, here's the patch for curious:

--- libatalk/adouble/ad_open.c.orig     2022-05-26 17:51:48.604464910 -0400
+++ libatalk/adouble/ad_open.c  2022-05-27 10:51:33.467798497 -0400
@@ -1574,6 +1574,8 @@
     uint32_t required_len;

        if (eid >= ADEID_MAX) {
+               LOG(log_error, logtype_ad, "ad_entry_check_size %d is greater than %d",
+                   eid, ADEID_MAX);
                return false;
        if (got_len == 0) {
@@ -1585,6 +1587,7 @@
                 * Shouldn't happen: implicitly initialized to zero because
                 * explicit initializer missing.
+               LOG(log_error, logtype_ad, "ad_entry_check_size explicit initializer missing");
                return false;
        if (ad_checks[eid].expected_len == -1) {
@@ -1594,6 +1597,8 @@
        if (ad_checks[eid].fixed_size) {
                if (ad_checks[eid].expected_len != got_len) {
                        /* Wrong size fo fixed size entry. */
+                       LOG(log_error, logtype_ad, "ad_entry_check_size wrong size to fixed size entry (%d != %d)",
+                           ad_checks[eid].expected_len, got_len);
                        return false;
         required_len = got_len;
@@ -1604,12 +1609,16 @@
                                 * Too small for variable sized entry with
                                 * minimum size.
+                               LOG(log_error, logtype_ad, "ad_entry_check_size too small for variable sized entry (%d < %d)",
+                                   got_len, ad_checks[eid].expected_len);
                                return false;
         required_len = got_len;
                } else {
                        if (got_len > ad_checks[eid].expected_len) {
                                /* Too big for variable sized entry. */
+                               LOG(log_error, logtype_ad, "ad_entry_check_size too big for variable sized entry (%d > %d)",
+                                   got_len, ad_checks[eid].expected_len);
                                return false;
@@ -1621,10 +1630,14 @@
        if (off + required_len < off) {
                /* wrap around */
+               LOG(log_error, logtype_ad, "ad_entry_check_size wrap around (%d + %d < %d)",
+                   off, required_len, off);
                return false;
        if (off + required_len > bufsize) {
                /* overflow */
+               LOG(log_error, logtype_ad, "ad_entry_check_size overflow (%d + %d > %d)",
+                   off, required_len, bufsize);
                return false;
        return true;
@@ -1639,10 +1652,7 @@

        valid = ad_entry_check_size(eid, bufsize, off, len);
        if (!valid) {
-               return NULL;
-       }
-       if (off == 0 || len == 0) {
+               LOG(log_error, logtype_ad, "ad_entry: not valid");
                return NULL;

jac_cbi commented on 2022-05-26 22:03 (UTC) (edited on 2022-05-26 22:07 (UTC) by jac_cbi)

Took the gist from here: (last hunk is what is relevant)

--- libatalk/adouble/ad_open.c.orig     2022-05-26 17:51:48.604464910 -0400
+++ libatalk/adouble/ad_open.c  2022-05-26 17:53:32.887798244 -0400
@@ -1639,10 +1639,7 @@

        valid = ad_entry_check_size(eid, bufsize, off, len);
        if (!valid) {
-               return NULL;
-       }
-       if (off == 0 || len == 0) {
+               LOG(log_error, logtype_ad, "ad_entry: not valid");
                return NULL;

EDIT: WIP. It built fine, and produced the log message, but it looks like I need to add more of the logging that the FreeBSD patch had

K4ib4 commented on 2022-05-21 15:41 (UTC)

Same problem on 3.1.13 - reverting back to 3.1.12-8 works fine.

krist commented on 2022-05-19 07:19 (UTC)

+1 on 3.1.13 crashing on login, successfully reverted back to 3.1.12-8 as well. It's already been reported to the netatalk bug tracker, too (

igno2k commented on 2022-05-14 17:39 (UTC)

Same here - reverting back to 3.1.12-8 works fine.

karog commented on 2022-05-09 16:39 (UTC) (edited on 2022-05-11 12:40 (UTC) by karog)

I upgraded from 3.1.12-8 (working) to 3.1.13-2 which installs and starts up ok on odroid n2 5.17.1-1-aarch64-ARCH. However when I try to do a Time Machine backup, it fails and I get the following stack trace in afpd.log:

May 11 08:35:09.701724 afpd[78059] {ad_open.c:794} (error:ad): ad_header_read_ea("/mnt/ssddata/tm"): invalid metadata EA

May 11 08:35:09.703791 afpd[78059] {ad_open.c:812} (error:ad): ad_header_read_ea("/mnt/ssddata/tm"): deleted invalid metadata EA

May 11 08:35:09.782380 afpd[78059] {ad_open.c:794} (error:ad): ad_header_read_ea("/mnt/ssddata/tm"): invalid metadata EA

May 11 08:35:09.782488 afpd[78059] {ad_open.c:812} (error:ad): ad_header_read_ea("/mnt/ssddata/tm"): deleted invalid metadata EA

May 11 08:35:09.783915 afpd[78059] {ad_open.c:794} (error:ad): ad_header_read_ea("/mnt/ssddata/tm/."): invalid metadata EA

May 11 08:35:09.783958 afpd[78059] {ad_open.c:812} (error:ad): ad_header_read_ea("/mnt/ssddata/tm/."): deleted invalid metadata EA

May 11 08:35:10.174750 afpd[78059] {fault.c:123} (severe:Default): ===============================================================

May 11 08:35:10.174826 afpd[78059] {fault.c:124} (severe:Default): INTERNAL ERROR: Signal 11 in pid 78059 (3.1.13)

May 11 08:35:10.174839 afpd[78059] {fault.c:125} (severe:Default): ===============================================================

May 11 08:35:10.177351 afpd[78059] {fault.c:96} (severe:Default): PANIC: internal error

May 11 08:35:10.177398 afpd[78059] {fault.c:97} (severe:Default): BACKTRACE: 12 stack frames:

May 11 08:35:10.177411 afpd[78059] {fault.c:103} (severe:Default): #0 /usr/lib/ [0xffffa0e8b2a4]

May 11 08:35:10.177434 afpd[78059] {fault.c:103} (severe:Default): #1 /usr/lib/ [0xffffa0e8b408]

May 11 08:35:10.177446 afpd[78059] {fault.c:103} (severe:Default): #2 [0xffffa11907dc]

May 11 08:35:10.177457 afpd[78059] {fault.c:103} (severe:Default): #3 /usr/lib/ [0xffffa0e68184]

May 11 08:35:10.177467 afpd[78059] {fault.c:103} (severe:Default): #4 /usr/lib/ [0xffffa0e68ec4]

May 11 08:35:10.177477 afpd[78059] {fault.c:103} (severe:Default): #5 /usr/bin/afpd(+0x2e714) [0xaaaae052e714]

May 11 08:35:10.177487 afpd[78059] {fault.c:103} (severe:Default): #6 /usr/bin/afpd(+0x2f2e0) [0xaaaae052f2e0]

May 11 08:35:10.177497 afpd[78059] {fault.c:103} (severe:Default): #7 /usr/bin/afpd(afp_over_dsi+0x36c) [0xaaaae050d80c]

May 11 08:35:10.177507 afpd[78059] {fault.c:103} (severe:Default): #8 /usr/bin/afpd(main+0x978) [0xaaaae050b838]

May 11 08:35:10.177516 afpd[78059] {fault.c:103} (severe:Default): #9 /usr/lib/ [0xffffa0cab8fc]

May 11 08:35:10.177526 afpd[78059] {fault.c:103} (severe:Default): #10 /usr/lib/ [0xffffa0cab9d4]

May 11 08:35:10.177536 afpd[78059] {fault.c:103} (severe:Default): #11 /usr/bin/afpd(_start+0x30) [0xaaaae050bc70]

May 11 08:35:32.062869 afpd[78082] {auth.c:569} (error:AFPDaemon): afp_disconnect: primary reconnect failed

May 11 08:35:33.120390 afpd[78082] {dsi_stream.c:503} (error:DSI): dsi_stream_read: len:0, unexpected EOF

May 11 08:35:53.403021 afpd[78087] {auth.c:569} (error:AFPDaemon): afp_disconnect: primary reconnect failed

May 11 08:35:53.624786 afpd[78087] {dsi_stream.c:503} (error:DSI): dsi_stream_read: len:0, unexpected EOF

I reverted and it is working again.

naumovitch commented on 2022-05-08 22:28 (UTC)

Netatalk 3.1.13 is available!

sardo commented on 2021-11-19 22:57 (UTC)

[forehead slap] thanks. works now.

a821 commented on 2021-11-19 22:26 (UTC)

@sardo: just recompile netatalk against the new libldap library.

sardo commented on 2021-11-19 21:50 (UTC)

netatalk fails after libldap upgrade libldap-2.4.59-2 -> libldap-2.6.0-2

netatalk: error while loading shared libraries: cannot open shared object file: No such file or directory

lrwxrwxrwx 1 root root 18 Nov 13 11:37 /lib/ -> lrwxrwxrwx 1 root root 10 Nov 13 11:37 /lib/ -> -rwxr-xr-x 1 root root 384368 Nov 13 11:37 /lib/*

downgrading libldap circumvents the issue.

lrwxrwxrwx 1 root root 21 Jul 18 15:59 /lib/ -> -rwxr-xr-x 1 root root 318528 Jul 18 15:59 /lib/ lrwxrwxrwx 1 root root 23 Jul 18 15:59 /lib/ -> -rwxr-xr-x 1 root root 343168 Jul 18 15:59 /lib/ lrwxrwxrwx 1 root root 23 Jul 18 15:59 /lib/ -> lrwxrwxrwx 1 root root 21 Jul 18 15:59 /lib/ -> lrwxrwxrwx 1 root root 10 Jul 18 15:59 /lib/ ->*

frankspace commented on 2021-07-22 14:12 (UTC) (edited on 2021-07-22 14:13 (UTC) by frankspace)

FYI, the Samba project is aware that tdb 1.4.4 is broken:

...and they have in fact already released a tdb version 1.4.5. I note that tdb has already been flagged out-of-date on Arch. I tested compiling and installing tdb 1.4.5 myself, and the previous version of netatalk compiles just fine afterwards (I also confirmed that it didn't compile with tdb 1.4.4).

I expect the official repos will provide the fixed version of tdb in reasonable order. Therefore, may I respectfully suggest that instead of reverting to embedded tdb, changing the PKGBUILD to require "tdb>=1.4.5"?

denn commented on 2021-07-22 12:54 (UTC)

tdb 1.4.4-1 seems to provide broken tdb.h, so I've reverted to using embedded tdb for now

charo21 commented on 2021-07-22 09:37 (UTC)

The 3.1.12-6 version can not find its own "tdb", may be, and make error.

There are lots of errors like "/usr/include/tdb.h:1002:9: error: expected ‘;’ before ‘int’ 1002 | PUBLIC int tdb_chainunlock_read(struct tdb_context *tdb, TDB_DATA key);"

I think it's due to the latest version of autoreconf, because there are many warning like "warning: source file '$(top_srcdir)/etc/afpd/afp_dsi.c' is in a subdirectory, but option 'subdir-objects' is disabled" and "autoconf input should be named '', not '' "

How can I figure it out

a821 commented on 2021-03-18 22:00 (UTC)

Patch for missing python2-dbus. It uses the new dbus-python.

codingcatgirl commented on 2020-11-28 16:49 (UTC)

This only compiles with if i add -I/usr/include/tirpc to CFLAGS

Also, most of the software actually doesn't compile. All the tools in the /bin/ directory aren't getting compiled and are missing in the resulting package, because the subdir-options option for Automake is missing (apparently a backwards-incompatible requirement in new Automake versions).

It would be nice if you could add a patch for this as well so all the tools that come with netatalk (megatron, afppasswd, …) are included in the package as well, as i'm missing them right now.

i1314925 commented on 2020-10-22 14:34 (UTC)

'cracklib' dependency is missing.

carlosb commented on 2020-10-02 14:54 (UTC)

Managed to build this on a RPI 4 8Gb with the aarch64 architecture. Just add aarch64 to the archlist or run makepkg -A to ignore the archictecture.

RadioGnome1971 commented on 2019-08-31 20:15 (UTC)

After installing the latest build, netatalk.service failed to start..

netatalk resulted in netatalk: error while loading shared libraries: cannot open shared object file: No such file or directory

Installing extra/mariadb-libs as a dependency solved this.

It is not mentioned in the PKGBUILD

jfburdet commented on 2019-08-20 08:53 (UTC) (edited on 2019-08-20 08:53 (UTC) by jfburdet)

There is a problem with libevent version.

A dirty temporary trick is to do :

cd /usr/lib

ln -s

neik commented on 2019-08-12 12:00 (UTC)


there seems to be a problem with the current version of libevent.

> netatalk

netatalk: error while loading shared libraries: cannot open shared object file: No such file or directory

gavinbeatty commented on 2019-06-30 16:51 (UTC) (edited on 2019-06-30 16:52 (UTC) by gavinbeatty)

I built this on Raspberry Pi 3 by adding aarch64 to the arch list, and it's handling Time Machine successfully for me.

zman0900 commented on 2019-03-21 02:22 (UTC) (edited on 2019-03-21 02:34 (UTC) by zman0900)

Looks like you can add "--without-mysql-config" config option to ignore any installed mysql. This seems to skip building the mysql backend the same as when building in a clean chroot where mysql lib is not installed.

Also this seems to be bundling its own "tdb" instead of linking to the system tdb. That can be fixed by added a dependency on tdb and config option "--without-tdb"

Also ACL support is being built automatically, so it should either be disabled with "--without-acls" or explicitly enabled with "--with-acls" and add the "acl" dependency.

jabbas commented on 2019-01-11 07:11 (UTC)

Latest version requires mysql-client package to be installed (it's linking with

claunia commented on 2017-07-29 15:53 (UTC)

Hello, After last archlinux glibc update it is no longer able to link: /usr/bin/ld: afppasswd.o: relocation R_X86_64_32S against `.rodata.str1.1' can not be used when making a shared object; recompile with -fPIC /usr/bin/ld: afppasswd.o: relocation R_X86_64_PC32 against symbol `lseek64@@GLIBC_2.2.5' can not be used when making a shared object; recompile with -fPIC /usr/bin/ld: final link failed: Bad value collect2: error: ld returned 1 exit status

f_kuehne commented on 2017-07-12 15:22 (UTC) (edited on 2017-07-12 15:24 (UTC) by f_kuehne)

Hello , i'm not able to build the current version appart from those warings: aclocal: warning: autoconf input should be named '', not '' automake: warning: autoconf input should be named '', not '' ... test/afpd/ but option 'subdir-objects' is disabled test/afpd/ warning: source file '$(top_srcdir)/etc/afpd/spotlight.c' is in a subdirectory, .. there are many of this kind the final error is access denied for ./confige at line 32 in ./PKGBUILD: 31: CFLAGS="-Wno-unused-result -O2" \ 32: ./configure --prefix=/usr --localstatedir=/var/state --sysconfd$ 33: --sbindir=/usr/bin --enable-silent-rules --with-init-style=syst$ 34: --with-cracklib --with-cnid-cdb-backend --enable-pgp-uam --with$ 35: make Because I wanted to solve it quick and dirtdy I changed the permissions for the dir and alle files to 777 but it didn't help. This is for a armv7 (RaspberryPi) install.

andreas_baumann commented on 2017-03-22 08:12 (UTC)

pkg-config was missing.

andreas_baumann commented on 2017-03-22 07:11 (UTC)

Doesn't build: libtoolize: Consider adding 'AC_CONFIG_MACRO_DIRS([macros])' to, libtoolize: and rerunning libtoolize and aclocal. macros/netatalk.m4:4: warning: underquoted definition of AX_CHECK_DOCBOOK macros/netatalk.m4:4: run info Automake 'Extending aclocal' macros/netatalk.m4:4: or see error: possibly undefined macro: AC_MSG_ERROR If this token and others are legitimate, please use m4_pattern_allow. See the Autoconf documentation. error: possibly undefined macro: AC_DEFINE configure:15977: error: possibly undefined macro: AC_MSG_WARN autoreconf: /usr/bin/autoconf failed with exit status: 1 ==> ERROR: A failure occurred in build(). Aborting...

zman0900 commented on 2016-12-04 01:21 (UTC) (edited on 2016-12-04 01:25 (UTC) by zman0900)

Ignore the rest of this comment. Turns out afpstats doesn't work unless you have "afpstats = yes" in the config file. <strike> Something seems to be messed up with the dbus related stuff installed by this package. The afpstats program doesn't work, even a root: ┌─[dan@nas]─[~] └──╼ sudo afpstats Traceback (most recent call last): File "/usr/bin/afpstats", line 16, in main "/org/netatalk/AFPStats") File "/usr/lib/python2.7/site-packages/dbus/", line 241, in get_object follow_name_owner_changes=follow_name_owner_changes) File "/usr/lib/python2.7/site-packages/dbus/", line 248, in __init__ self._named_service = conn.activate_name_owner(bus_name) File "/usr/lib/python2.7/site-packages/dbus/", line 180, in activate_name_owner self.start_service_by_name(bus_name) File "/usr/lib/python2.7/site-packages/dbus/", line 278, in start_service_by_name 'su', (bus_name, flags))) File "/usr/lib/python2.7/site-packages/dbus/", line 651, in call_blocking message, timeout) DBusException: org.freedesktop.DBus.Error.ServiceUnknown: The name org.netatalk.AFPStats was not provided by any .service files ┌─[✗]─[dan@nas]─[~] └──╼ pacman -Qo /etc/dbus-1/system.d/netatalk-dbus.conf /etc/dbus-1/system.d/netatalk-dbus.conf is owned by netatalk 3.1.10-1 ┌─[dan@nas]─[~] └──╼ grep 'org.netatalk.AFPStats' /etc/dbus-1/system.d/netatalk-dbus.conf <allow own="org.netatalk.AFPStats"/> <allow send_destination="org.netatalk.AFPStats"/> <allow receive_sender="org.netatalk.AFPStats"/> </strike>

denn commented on 2016-09-21 15:47 (UTC)

Updated to 3.1.10

maclinuxfree commented on 2016-07-23 21:32 (UTC)

pls update to 3.1.9 thank you

denn commented on 2016-01-02 21:18 (UTC)

Updated to 3.1.8

fuerst commented on 2015-12-03 13:19 (UTC)

@lenovomi: Do you encounter the problems since you start using TM at your Arm device? Sounds a bit like corrupted hardware. Some things to check: * "dmesg" for any hard disk related errors. * "smartctl -A /dev/sda". "sda" is first drive, "sdb" second and so on. * Finally do a filesystem check including badblocks: "fsck -vcck /dev/<device-PARTITION>". See for details.

lenovomi commented on 2015-11-10 18:40 (UTC)

Guys, I have constantly problems with TM on my arch arm device. Sometimes my old backup got lost, sometimes whole TM archive got corrupted. I have sparse bundle as a storage for TM that lays on my my ext4. My config is almost identical to QNAP one. Any idea why those issues are constantly happening to me? Thanks.

justinkb commented on 2014-12-08 18:25 (UTC)

3.1.7 has been released, it seems.

rainrain commented on 2014-08-29 22:56 (UTC)

Netatalk 3.1.6 is available!

mikejuni commented on 2014-08-14 08:41 (UTC)

3.1.5 is out.

trialuser commented on 2014-07-31 19:06 (UTC)

Reinstalling acl fixed it for me, thanks for the help denn!

trialuser commented on 2014-07-30 17:58 (UTC)

failed with: Makefile:631: recipe for target 'libutil_la-unix.lo' failed make[3]: *** [libutil_la-unix.lo] Error 1 make[3]: *** Waiting for unfinished jobs.... make[3]: Leaving directory '/tmp/makepkg/netatalk/src/netatalk-3.1.3/libatalk/util' Makefile:674: recipe for target 'all-recursive' failed make[2]: *** [all-recursive] Error 1 make[2]: Leaving directory '/tmp/makepkg/netatalk/src/netatalk-3.1.3/libatalk' Makefile:512: recipe for target 'all-recursive' failed make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory '/tmp/makepkg/netatalk/src/netatalk-3.1.3' Makefile:443: recipe for target 'all' failed make: *** [all] Error 2 ==> ERROR: A failure occurred in build(). Aborting... ==> ERROR: Makepkg was unable to build netatalk. any ideas?

nariox commented on 2014-07-23 15:37 (UTC)

Could you add 'armv7h' to the arch? It builds and works well for me if I edit the PKGBUILD file.

denn commented on 2014-06-08 19:26 (UTC)

Updated to 3.1.2 and patched to use avahi-client instead of dns_sd

mikejuni commented on 2014-06-08 14:02 (UTC)

3.1.2 is out.

mikejuni commented on 2014-05-04 12:38 (UTC)

It seems like netatalk is currently linked against dns_sd but not avahi-client. It is still usable but I"m not sure if it is optimal.

SJ_UnderWater commented on 2014-02-22 19:24 (UTC)

I have disowned the package

SJ_UnderWater commented on 2014-02-18 14:58 (UTC)

libtool is a member of base-devel, please follow instructions

cxcv commented on 2014-02-17 20:34 (UTC)

missing dependency: libtool

SJ_UnderWater commented on 2014-02-16 15:55 (UTC)

Precisely because it fixed the issues users were having, check the the changelog. It also was an update, from 3.0.6 to 3.0.7

freswa commented on 2014-02-16 15:48 (UTC)

Why "update" to old stable? See on left side. 3.1.0 is new stable... Please update.

FalconGER commented on 2014-02-07 19:36 (UTC)

source-link does not work :-/ I tried successfully :) simply changed in PKGBUILD

marmotta commented on 2014-02-05 18:33 (UTC)

I have edited pkgbuild to install (new???) version 3.0.7 Same dbus error in compile, but now TimeMachine work perfectly!!!

SJ_UnderWater commented on 2014-01-26 21:37 (UTC)

I will be recusing myself soon as package maintainer, since making some lifestyle changes which mean I won't be using netatalk anymore (running Server on a spare machine instead). The package itself has been fairly stable since the 2.2 days, but with Python, the newer Spotlight support, and Autoconf warnings piling up, its up to someone new to decide what the default featureset should be, and how to set it up. mDNSResponder (in the AUR) remains a possibility (and fewer Avahi API warnings), but Tracker (for Spotlight) appears to require additional headers which aren't compiled into the existing package. I hope someone will step up before I orphan the package.

moonman commented on 2014-01-21 10:09 (UTC)

Just an FYI and for your consideration: you can bump the pkgrel so the package gets picked up by tools like yaourt for automatic update. As the maintainer you are probably informed better of when it needs to be rebuilt.

Butcho commented on 2014-01-19 22:38 (UTC)

Indeed the error goes away but ever since the libgcrypt update, I get a (null) when I search for timemachine drives, I get a dsi_stream_read: len:0, unexpected EOF Jan 19 17:37:48.384358 afpd[1507] {dsi_tcp.c:241} (I:DSI): AFP/TCP session from Jan 19 17:37:48.417518 afpd[1230] {main.c:151} (I:AFPDaemon): child[1506]: exited 4 Jan 19 17:37:59.243259 afpd[1516] {dsi_tcp.c:241} (I:DSI): AFP/TCP session from Jan 19 17:37:59.245355 afpd[1516] {uams_dhx2_pam.c:329} (I:UAMS): DHX2 login: timemachine Jan 19 17:37:59.345926 afpd[1516] {uams_dhx2_pam.c:214} (I:UAMS): PAM DHX2: PAM Success Jan 19 17:37:59.362695 afpd[1516] {uams_dhx2_pam.c:709} (I:UAMS): DHX2: PAM Auth OK! Jan 19 17:37:59.362717 afpd[1516] {auth.c:232} (N:AFPDaemon): AFP3.3 Login by timemachine Jan 19 17:37:59.375302 afpd[1516] {auth.c:828} (N:AFPDaemon): AFP logout by timemachine Jan 19 17:37:59.376437 afpd[1516] {dsi_stream.c:504} (E:DSI): dsi_stream_read: len:0, unexpected EOF Jan 19 17:37:59.376508 afpd[1516] {afp_dsi.c:503} (N:AFPDaemon): afp_over_dsi: client logged out, terminating DSI session Jan 19 17:37:59.377023 afpd[1516] {afp_dsi.c:106} (N:AFPDaemon): AFP statistics: 0.60 KB read, 0.45 KB written Jan 19 17:37:59.377070 afpd[1516] {dircache.c:615} (I:AFPDaemon): dircache statistics: entries: 0, lookups: 0, hits: 0, misses: 0, added: 0, removed: 0, expunged: 0, evicted: 0 Jan 19 17:37:59.378417 afpd[1230] {main.c:153} (I:AFPDaemon): child[1516]: done

SJ_UnderWater commented on 2014-01-19 21:23 (UTC)

When a library update breaks the package, you must recompile it yourself; there is no change necessary to what I've posted here, which is a source package. I recompiled netatalk from the tarball and successfully completed a TimeMachine backup. I will unflag the package tomorrow. I will also look into making python2 the requirement. mdnsresponder is in the AUR, not the main repository.

Butcho commented on 2014-01-19 19:17 (UTC)

libgcrypt was updated and broke netatalk. Please update.

jakethedog commented on 2014-01-18 08:52 (UTC)

@SJ_UnderWater: There is a problem with the new libgcrypt 1.6, because of missing "/usr/lib/" could you recompile this package. Thanks

jakethedog commented on 2014-01-18 08:10 (UTC)

The problem with afpstats is related to the incompatibly of the afpstats script with python3. Just change the first line from "#!/usr/bin/env python" to "#!/usr/bin/env python2". Python2 should be installed of course.

marmotta commented on 2014-01-08 06:41 (UTC)

Sorry for my poor english... I wanted to say this: If i write: "pacman -S mdnsresponder" the packagemanager download avahi. You do not have this problem with TimeMachine adisk service?

SJ_UnderWater commented on 2014-01-08 02:02 (UTC)

you can't have mdnsresponder and avahi, they both do the same thing.

marmotta commented on 2014-01-07 18:27 (UTC)

Any news? It is not possible that only I have this problem...

marmotta commented on 2014-01-05 10:25 (UTC)

My first install is in real machine, fresh install in VM is only for exclude other problems. mdnsresponder is already installed from pacman and install avahi. You want info log? Is very long, where i post it?

SJ_UnderWater commented on 2014-01-05 06:09 (UTC)

it does look like it could be a compilation issue, but the file in question afp_avahi.c hasn't changed since 2012. The (null) points to the printf() in that file, but it's not obvious why. It may not be dbus after all, but your use of a VM does raise questions. I would suggest using the INFO logging level and checking for the Bonjour entries. If the log line is correct, something is going wrong memory-wise. You may also want to try building mdnsresponder and compiling against that instead, if Avahi continues to give you problems. At worst you could create the service entry yourself.

marmotta commented on 2014-01-04 18:28 (UTC)

I have compiled in VM with fresh install of updated Arch. The problem remains the same.

marmotta commented on 2014-01-03 21:29 (UTC)

Is true..... My dbus is updated (last day) and running without a errors, how i fix netatalk?

SJ_UnderWater commented on 2014-01-03 21:23 (UTC)

afpstats also uses dbus, the documentation says it clearly

marmotta commented on 2014-01-03 20:58 (UTC)

Or python? [marmotta@Media system.d]$ afpstats File "/usr/bin/afpstats", line 26 print name ^ SyntaxError: invalid syntax

SJ_UnderWater commented on 2014-01-03 20:46 (UTC)

if dbus is failing to connect to systemd during make, and avahi and afpd aren't communicating well during startup, then it does seem like dbus is the issue.

marmotta commented on 2014-01-03 20:40 (UTC)

systemd is ok, my system is full systemd and updated. My afp.conf is correct: Jan 03 17:18:09.757480 netatalk[19101] {netatalk_conf.c:582} (D5:AFPDaemon): createvol(volume: 'TimeMachine', path: "/home/marmotta/MediaServer/TimeMachine", preset: '-'): BEGIN Jan 03 17:18:09.757574 netatalk[19101] {netatalk_conf.c:256} (D5:AFPDaemon): Volume "/home/marmotta/MediaServer/TimeMachine" ACL support: yes Jan 03 17:18:09.757692 netatalk[19101] {vfs.c:852} (D5:AFPDaemon): initvol_vfs: enabling EA support with native EAs Jan 03 17:18:09.757751 netatalk[19101] {netatalk_conf.c:115} (D5:AFPDaemon): get_uuid('TimeMachine'): UUID: 'B4F77609-B6FF-DA9F-0909-26FDDDF4F089' Jan 03 17:18:09.757796 netatalk[19101] {netatalk_conf.c:956} (D5:AFPDaemon): Volume 'TimeMachine': UUID 'B4F77609-B6FF-DA9F-0909-26FDDDF4F089' Jan 03 17:18:09.757826 netatalk[19101] {netatalk_conf.c:965} (D5:AFPDaemon): createvol: END: 0 adisk service approximately a month ago it was ok, now i have this problem. Incompatibility with updated avahi? dbus?

SJ_UnderWater commented on 2014-01-03 18:16 (UTC)

make sure systemd is working or the install-data-hook will fail. Avahi is working, because you have an adisk service entry. Make sure you configured your shares correctly, see and the original Netatalk documentation at the upstream URL above.

marmotta commented on 2014-01-03 17:56 (UTC)

I have a error when make a package: Failed to get D-Bus connection: Connection terminated during authentication. make[4]: [install-data-hook] Errore 1 (ignorato) Avahi discovery not work, my mac show only a time machine "null" drive: [marmotta@Media netatalk]$ avahi-browse -ar + eth0 IPv4 Media _device-info._tcp local + eth0 IPv4 Media Apple TimeMachine local + eth0 IPv4 Media Apple File Sharing local + eth0 IPv4 MEDIA(SMB) Rete Microsoft Windows local + eth0 IPv4 Media [**:**:**:**:**:**] Macchine local + eth0 IPv4 Media SSH Terminale remoto local + eth0 IPv4 Media SFTP File Transfer local = eth0 IPv4 Media _device-info._tcp local hostname = [Media.local] address = [] port = [9] txt = ["model=RackMac"] = eth0 IPv4 Media Apple TimeMachine local hostname = [Media.local] address = [] port = [548] txt = ["dk0=adVN=(null),adVF=0xa1,adVU=(null)" "sys=waMa=0,adVF=0x100"] = eth0 IPv4 Media Apple File Sharing local hostname = [Media.local] address = [] port = [548] txt = [] The netatalk log not show errors

SJ_UnderWater commented on 2013-12-17 01:54 (UTC)

Namcap says (and I agree), that python is required. You're free to edit the PKGBUILD, but it's not like you can half-run afpstats without it.

richjn commented on 2013-12-17 00:17 (UTC)

Ahh, I see Thanks for the explanation for libevent. Could python be an optdepends (to run afpstats) or is it required to build afpstats in the first place ? Richard

SJ_UnderWater commented on 2013-12-16 04:15 (UTC)

The idea is to use the pkg repository's libevent, not the included libevent, hence the dependency. As for python, it's required by the afpstats script.

richjn commented on 2013-12-15 20:37 (UTC)

Hi Thanks for the package, I just have a query / observation. 1) libevent is shown as a dependancy but the PKGBUILD has --with-libevent=no I tried without libevent installed and compile failed. It seems the normal --with-"option"=no convention does not work here where a --disable-libevent does and compile then works fine without libevent. 2) I probably haven't researched enough so excuse me if I'm missing the obvious but why the need for python ? it compiles and runs fine here without python. Regards Richard

freswa commented on 2013-07-17 14:33 (UTC)

Just had to restart whole server because of kernelchanges. It is working now - strange since I already restarted the services...

SJ_UnderWater commented on 2013-07-17 03:57 (UTC)

I suggest you check to make sure any mDNS service can be broadcast first, check the avahi documentation. Otherwise you will have to increase the logging level and look for errors. The hostname shouldn't be added as a prefix to the service name, that's incorrect, besides the fact that service names must begin with an underscore "_"

freswa commented on 2013-07-16 12:43 (UTC)

heimdall is the hostname of that server. But don't know where the _ is coming from...? Made all settings following the Wikipage.

SJ_UnderWater commented on 2013-07-16 05:11 (UTC)

how are you getting the heimdall_ prefix? that doesn't look correct. did you try turning up the verbosity to see where it's coming from?

freswa commented on 2013-07-15 14:42 (UTC)

Cant connect to netatalk while getting this log entries: Jul 15 16:34:41.792684 afpd[26005] {afp_mdns.c:92} (E:AFPDaemon): Failed to register mDNS service: heimdall_afpovertcp._tcp.: code=-65537 message repeated 2 times Jul 15 16:34:41.792769 afpd[26005] {afp_mdns.c:92} (E:AFPDaemon): Failed to register mDNS service: heimdall_device-info._tcp.: code=-65537 message repeated 2 times Help appreciated

SJ_UnderWater commented on 2013-06-28 18:58 (UTC)

pkg-config is a memeber of base-devel, see

commented on 2013-06-28 11:08 (UTC)

Please add `pkg-config` to the dependencies. If that package is not installed, autoconf will fail building the appropriate files.

commented on 2013-06-05 11:22 (UTC)

sorry, my mistake, I compiled it while I had tcp_wrappers installed, and after the last update intervention on arch I removed it and didnt re-install. great job by the way, works like a charm usually :)

SJ_UnderWater commented on 2013-06-04 20:57 (UTC)

there is no explicit tcp_wrappers dependency, the website itself lists it under "optional". It was explicitly included in earlier versions of this package, but was removed when it dropped from Extra to Unsupported.

commented on 2013-06-04 20:41 (UTC)

so is it just me or is this thing dependent of the tcp_wrappers package?

SJ_UnderWater commented on 2013-06-04 18:32 (UTC)

the description pre-dates my involvement as the maintainer, I've kept it. if by build flags you mean CFLAGS, -Wno-unused-result cleans up the compilation warnings, and -O2 is a recent addition to prevent another "# warning _FORTIFY_SOURCE requires compiling with optimization (-O)"

fauno commented on 2013-06-04 14:25 (UTC)

hey, could you explain what the build flags for? it works ok without them though i haven't done intensive testing. here's a modified version of your pkgbuild: (also the description doesn't seem right "kernel level"?)

SJ_UnderWater commented on 2013-06-03 19:48 (UTC)

thank you for calling my attention to the usrbin changes, new version up.

WonderWoofy commented on 2013-06-03 16:16 (UTC)

All that you need to do is add "--sbindir=/usr/bin" to the ./configure line to make it compatible with the /usr/bin move.

commented on 2013-06-03 12:58 (UTC)

In regards to the latest update: It appears that the netatalk binary is installed to /usr/sbin. It should now be installed to /usr/bin.

commented on 2013-04-19 08:36 (UTC)

Can you please add --enable-krbV-uam to configure options? It's needed to get netatalk work with kerberos.

SJ_UnderWater commented on 2013-03-27 09:57 (UTC)

updated to 3.0.3, unfortunately this adds a python dependency for `afpstates` using a fully-updated installation and a fresh copy of the tarball, the `cp` command is necessary.

Fraterius commented on 2013-03-26 12:52 (UTC)

I'm getting build error with: cp: will not overwrite just-created './missing' with '/usr/share/automake-1.13/missing' It's building ok without cp command though. Is it necessary in the first place?

SJ_UnderWater commented on 2013-02-14 04:28 (UTC)

i cannot find any reference to "freedesktop.avahi" in my src/ or pkg/ directories, though I don't use systemd. As far as I can tell, it is your responsibility to ensure Avahi is loaded before netatalk starts (like sysvinit); you could also modify the netatalk.service dependencies.

commented on 2013-02-13 07:01 (UTC)

I had to run the following commands to enable bonjour discovery (shows up in the sidebar) on my mac. I could still connect to the server without it by manually entering the ip address: systemctl enable avahi-daemon.socket systemctl enable avahi-daemon.service The installation seems to target the wrong file, the correct files are above. I found an error in journalctl and the service "status dbus-org.freedesktop.Avahi.service" had an error starting. Running this showed the cause: systemctl status dbus-org.freedesktop.Avahi.service dbus-org.freedesktop.Avahi.service Loaded: error (Reason: No such file or directory) Active: inactive (dead)

SJ_UnderWater commented on 2013-01-05 09:50 (UTC)

should be fixed now, but there's more going on with the deprecated macros. also a minor install script fix

Kev commented on 2013-01-04 19:17 (UTC)

Does not compile: -> Fixing... error: 'AM_CONFIG_HEADER': this macro is obsolete. You should use the 'AC_CONFIG_HEADERS' macro instead. /usr/share/aclocal-1.13/obsolete-err.m4:12: AM_CONFIG_HEADER is expanded from... the top level autom4te: /usr/bin/m4 failed with exit status: 1 aclocal: error: echo failed with exit status: 1 autoreconf: aclocal failed with exit status: 1 ==> FEHLER: Ein Fehler geschah in build().

SJ_UnderWater commented on 2012-12-07 10:08 (UTC)

wiki article has been updated

h50 commented on 2012-10-17 01:49 (UTC)

I fixed it. There seemed to be a corrupt .DS_Store file in there. I just deleted it.

h50 commented on 2012-10-17 01:35 (UTC)

More verbose logging: Oct 16 21:34:42 h5pi cnid_metad[5319]: Set syslog logging to level: DEBUG9 Oct 16 21:34:42 h5pi cnid_metad[5319]: CNID Server: localhost:4700 Oct 16 21:34:42 h5pi cnid_metad[5319]: Global unix charset is UTF8 Oct 16 21:34:42 h5pi cnid_metad[5319]: Global vol charset is UTF8 Oct 16 21:34:42 h5pi cnid_metad[5319]: Global mac charset is MAC_ROMAN Oct 16 21:34:42 h5pi cnid_metad[5319]: load_volumes: BEGIN Oct 16 21:34:42 h5pi cnid_metad[5319]: load_volumes: no volumes yet Oct 16 21:34:42 h5pi cnid_metad[5319]: load_volumes: loading: /etc/afp.conf Oct 16 21:34:42 h5pi cnid_metad[5319]: readvolfile: BEGIN Oct 16 21:34:42 h5pi cnid_metad[5319]: readvolfile: sections: 6 Oct 16 21:34:42 h5pi cnid_metad[5319]: createvol(volume: 'TimeMachine', path: "/mnt/raid/timeMachine", preset: '-'): BEGIN Oct 16 21:34:42 h5pi netatalk[5233]: Got SIGCHLD event Oct 16 21:34:42 h5pi netatalk[5233]: child[5319]: killed by signal 11 Oct 16 21:34:43 h5pi netatalk[5233]: Restarting 'cnid_metad' (restarts: 76)

h50 commented on 2012-10-17 01:30 (UTC)

Since I updated, I get continuous messages about cnid restarting, and the package just doesn't work anymore! Name : netatalk Version : 3.0.1-3 Oct 16 21:25:25 h5pi systemd[1]: Starting Netatalk AFP fileserver for Macintosh clients... Oct 16 21:25:25 h5pi systemd[1]: Started Netatalk AFP fileserver for Macintosh c lients. Oct 16 21:25:25 h5pi netatalk[4718]: Netatalk AFP server starting Oct 16 21:25:26 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 1) Oct 16 21:25:27 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 2) Oct 16 21:25:28 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 3) Oct 16 21:25:29 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 4) Oct 16 21:25:30 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 5) Oct 16 21:25:31 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 6) Oct 16 21:25:32 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 7) Oct 16 21:25:33 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 8) Oct 16 21:25:34 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 9) Oct 16 21:25:35 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 10) Oct 16 21:25:36 h5pi netatalk[4718]: Restarting 'cnid_metad' (restarts: 11) Oct 16 21:25:37 h5pi afpd[4719]: Netatalk AFP/TCP listening on

commented on 2012-10-16 17:05 (UTC)

Thanks. In particular in light of the "Add "netatalk" to your DAEMONS list in /etc/rc.conf" message I got upon update, I presumed that it should run with initd. The script can be found at

SJ_UnderWater commented on 2012-10-16 16:16 (UTC)

until I can reliably determine init/systemd at buildtime, this pkg is systemd-only. Adding back the rc.d/init portion is easy, just use the initscript template from the wiki, setting the DAEMON to netatalk. There are no special considerations there.

commented on 2012-10-16 14:56 (UTC)

The current version does not install anything in /etc/rc.d/netatalk/* and consequently a daemon cannot get started with from /etc/rc.conf. From what I understand, you intend to be compatible with both systemd and initd, but this is currently not working for initd.

SJ_UnderWater commented on 2012-10-08 22:56 (UTC)

another couple of fixes. I kept the systemd line, but moved from /lib to /usr/lib to prevent namcap errors. also includes a smarter install script that checks for rc.d first. I'm hoping to find a way to detect systemd/init at buildtime, but haven't found a good way yet. This way the package will work in both cases.

pac commented on 2012-10-08 21:13 (UTC)

Thanks for the update (especially for the moved state information). As I use a pure systemd init, I get the following (harmless) info during install: grep: /etc/rc.conf: No such file or directory >>> Add "netatalk" to your DAEMONS list in /etc/rc.conf >>> Additionally, I think the following fix in the PKGBUILD is not needed: sed -i 's:/lib/systemd:/etc/systemd:' distrib/initscripts/Makefile.{am,in} systemd expects service files that are installed by a package management system to reside in /usr/lib/systemd, whereas /etc/systemd is intended for manual overrides of these service definitions.

SJ_UnderWater commented on 2012-10-05 13:46 (UTC)

removed rc.d reference in the install file

commented on 2012-10-05 13:00 (UTC)

Thanks for the update, it solves some network issues from the previous version. I've juste noticed there is still an issue when your uninstall the package with pacman -Rns : "/tmp/alpm_vj4BiN/.INSTALL: line 22: rc.d: command not found" ! It must be related to the old rc.d !

SJ_UnderWater commented on 2012-10-05 04:12 (UTC)

now using version 3.0.1, systemd, and moved state information to /var/state/netatalk. I hope this helps others with FHS compliance and those with versioned /etc directories. I'm not on systemd yet, so please tell me if there is an issue

fauno commented on 2012-09-20 17:48 (UTC)

Added a modified version to Parabola's [social] repo, thanks for your work!

emphire commented on 2012-09-03 00:40 (UTC)

Please add "--with-init-style=systemd" to configure flags so a systemd service file is generated. Thanks!

SJ_UnderWater commented on 2012-08-08 08:54 (UTC)

glad it was fixed. i agree that the error messages aren't very useful, but the "CNID DB" messages are usually related to permissions, i.e. it couldn't create the db.

munkoil commented on 2012-08-08 08:53 (UTC)

Reboot on both server and OS X solved the problem.

munkoil commented on 2012-08-08 08:31 (UTC)

BTW, I don't get any details in syslog.

munkoil commented on 2012-08-08 08:22 (UTC)

I'm having a problem with netatalk 3. When I login in to a share no files are shown for a while, then they show up with the following error message: Something wrong with the volume's CNID DB, using temporary CNID DB instead.Check server messages for details. Switching to read-only mode. The thing is that I had no database for the volume yet. I tried both with and without running: dbd -r /path/to/share If I share home directories with [homes] I do not receive this error and everything works as expected. Any ideas? ; ; Netatalk 3.x configuration file ; [Global] ; Global server settings mimic model = RackMac [Homes] basedir regex = /home [media] path = /scratch/media/

commented on 2012-07-23 08:10 (UTC)

I replaced binutils-multilib gcc-multilib libtool-multilib with binutils gcc libtool and now it works :)

SJ_UnderWater commented on 2012-07-18 01:48 (UTC)

i can only assume it's caused by multilib/lib32-libltdl, but could you make sure you have all of base and base-devel updated?

commented on 2012-07-17 08:33 (UTC)

I got the following error when upgrade error: possibly undefined macro: AC_LTDL_DLLIB If this token and others are legitimate, please use m4_pattern_allow. See the Autoconf documentation. error: possibly undefined macro: AC_LTDL_DLSYM_USCORE autoreconf: /usr/bin/autoconf failed with exit status: 1 but I have core/libltdl and multilib/lib32-libltdl installed on my system

SJ_UnderWater commented on 2012-07-15 18:56 (UTC)

the problem is that /etc/netatalk/*.conf was the location for some important, generated, permanent files. I say permanent because they normally survive package removal, and since they contain UUIDs for the machine (and any TM volumes) they really should hang around in case the machine ever runs netatalk again. I agree completely about the variable/permanent issue, but there are continuity issues here as well as FHS. Really the netatalk people should never have decided to merge those trees, but I've decided to choose continuity over strict FHS compliance, at least for the near future.

pac commented on 2012-07-15 14:12 (UTC)

@SJ_UnderWater: Personally, I don't like having state information in /etc (and the FHS doesn't support this either). Also, there are people that use a version control system for /etc which will completely break if there is state information in it. Don't you think /var/lib might be an appropriate place, as it's explicitly declared to hold "variable state information"?

SJ_UnderWater commented on 2012-07-15 03:42 (UTC)

after some discussion, and for historical reasons, localstatedir (previously /var) has been relocated from the recent /opt to /etc, to ensure that the afp_*.conf files are in the same place as 2.2.x. Since 3.0 creates ${localstatedir}/netatalk, /var is not an option (FHS rules), /opt seemed like the appropriate place because the dir will contain both temp and some permanent data (uuids), but this makes a break with previous installations. Despite the creation of temp files in /etc, i will be configuring the package this way from now on.

SJ_UnderWater commented on 2012-07-12 09:41 (UTC)

so i've disabled the bundled version of libevent, and added it as a dependency instead of python2. Now its up to the user if they wish to install python, which is a optional dependency of the libevent pkg.

munkoil commented on 2012-07-12 06:02 (UTC)

Hi, I get the following error when upgrading. I also tried to uninstall netatalk-2.2.3 first and then install netatalk-3. Still same problem. Any thoughts? ... error: failed to commit transaction (conflicting files) netatalk: /usr/bin/ exists in filesystem netatalk: /usr/include/evdns.h exists in filesystem netatalk: /usr/include/event.h exists in filesystem netatalk: /usr/include/event2/buffer.h exists in filesystem netatalk: /usr/include/event2/buffer_compat.h exists in filesystem netatalk: /usr/include/event2/bufferevent.h exists in filesystem netatalk: /usr/include/event2/bufferevent_compat.h exists in filesystem netatalk: /usr/include/event2/bufferevent_ssl.h exists in filesystem netatalk: /usr/include/event2/bufferevent_struct.h exists in filesystem netatalk: /usr/include/event2/dns.h exists in filesystem netatalk: /usr/include/event2/dns_compat.h exists in filesystem netatalk: /usr/include/event2/dns_struct.h exists in filesystem netatalk: /usr/include/event2/event-config.h exists in filesystem netatalk: /usr/include/event2/event.h exists in filesystem netatalk: /usr/include/event2/event_compat.h exists in filesystem netatalk: /usr/include/event2/event_struct.h exists in filesystem netatalk: /usr/include/event2/http.h exists in filesystem netatalk: /usr/include/event2/http_compat.h exists in filesystem netatalk: /usr/include/event2/http_struct.h exists in filesystem netatalk: /usr/include/event2/keyvalq_struct.h exists in filesystem netatalk: /usr/include/event2/listener.h exists in filesystem netatalk: /usr/include/event2/rpc.h exists in filesystem netatalk: /usr/include/event2/rpc_compat.h exists in filesystem netatalk: /usr/include/event2/rpc_struct.h exists in filesystem netatalk: /usr/include/event2/tag.h exists in filesystem netatalk: /usr/include/event2/tag_compat.h exists in filesystem netatalk: /usr/include/event2/thread.h exists in filesystem netatalk: /usr/include/event2/util.h exists in filesystem netatalk: /usr/include/evhttp.h exists in filesystem netatalk: /usr/include/evrpc.h exists in filesystem netatalk: /usr/include/evutil.h exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/libevent.a exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/libevent_core.a exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/libevent_extra.a exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/libevent_openssl.a exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/libevent_pthreads.a exists in filesystem netatalk: /usr/lib/ exists in filesystem netatalk: /usr/lib/pkgconfig/libevent.pc exists in filesystem netatalk: /usr/lib/pkgconfig/libevent_openssl.pc exists in filesystem netatalk: /usr/lib/pkgconfig/libevent_pthreads.pc exists in filesystem Errors occurred, no packages were upgraded.

SJ_UnderWater commented on 2012-07-11 23:06 (UTC)

ok, so a new release here, which discards DDP (see netatalk-ddp for those features). Python2 had to be added because of, which i will try to patch in the future to Py3. The remaining fix is to prevent ldconfig.

commented on 2012-06-17 21:44 (UTC)

Thanks, that worked, not sure how base-devel got out of date.

SJ_UnderWater commented on 2012-06-16 18:10 (UTC)

make sure you have all of base-devel installed and updated. just rebuilt it on my machine w/o errors.

commented on 2012-06-16 05:21 (UTC)

I haven't looked into this much but I'm getting and error: error: possibly undefined macro: AC_LTDL_DLLIB If this token and others are legitimate, please use m4_pattern_allow. See the Autoconf documentation. error: possibly undefined macro: AC_LTDL_DLSYM_USCORE autoreconf: /usr/bin/autoconf failed with exit status: 1

SJ_UnderWater commented on 2012-06-05 00:19 (UTC)

fixed, sorry for the trouble

emphire commented on 2012-06-04 22:32 (UTC)

I'm getting an error when updating: error: failed to commit transaction (conflicting files) netatalk: /usr/include/netatalk/at.h exists in filesystem $ pacman -Qo /usr/include/netatalk/at.h /usr/include/netatalk/at.h is owned by glibc 2.15-10

SJ_UnderWater commented on 2012-06-02 06:21 (UTC)

ok, i decided to keep the version number while making the changes, instead of making it 2.2.4dev (at least for now). I'll keep a closer eye on product-2-2 for additional fixes.

bsjeon commented on 2012-06-02 06:01 (UTC)

We neeed a patch, and autoreconf. "Fix missing UAM links":;a=commitdiff;h=61eb0f4ddde2c15ca16ab9aaabce3aa5927b8e05;hp=02b860940d3e886ac4eedb7194d7453018ec0bf4

SJ_UnderWater commented on 2012-05-02 19:29 (UTC)

it's more correct to say that the "missing" netatalk-ddp package is _unexpected_. It shouldn't fail on a missing package - that's probably unnecessary.

doits commented on 2012-05-02 10:55 (UTC)

so this basically means there is a bug in "makepkg -i", which wants to install all package-versions even if not specified? I'll report it as an makepkg bug then.

SJ_UnderWater commented on 2012-05-01 19:14 (UTC)

doits, that is in fact exactly what _does_ happen. With no options it builds netatalk alone, and will package it correctly. As far as I can tell the only issue is if you specify "makepkg -i" that it can't find netatalk-ddp and may fail. Netatalk is still built and packaged in the meantime.

doits commented on 2012-05-01 14:22 (UTC)

My problem was, I simply wanted to install netatalk. It is nowhere written I *must* set the "pkg"-flag to either "netatalk" or "netatalk-ddp". Without the flag it does not build and install at all. If you think that is right, it should at least gracefully fail before doing anything and tell the user that the flag is missing, so he knows whats up. I'd prefer to have the standard set up to build the simple "netatalk" (with simple "makepkg"), and you can use "makepkg --pkg netatalk-ddp" if you want the ddp-version.

SJ_UnderWater commented on 2012-04-26 23:43 (UTC)

i'm well aware of the standards, but if you read the wiki page, there is some discussion of why it works this way right now. I agree that it is a little hack-y and not perfect, but for maintaining what will be two different packages, this is the simplest way of making sure that both are kept in-line. After all, they're technically the same software package though the uses are very different. again: if you don't specify `--pkg netatalk-ddp`, then that package is not built.

doits commented on 2012-04-26 21:18 (UTC)

>SJ how do you install this? I tried with "makepkg -si", and indeed, it does not work. Since that is the standard way of how arch packages are built (, netatalk should work with it, too. Can you look at this?

SJ_UnderWater commented on 2012-04-26 18:27 (UTC)

if anyone has any suggestions, before 3.0 comes out and i officially branch the packages, let me know.

SJ_UnderWater commented on 2012-04-24 18:14 (UTC)

there may not be a way around that, but why is it looking for "any"? that alone seems wrong.

doits commented on 2012-04-24 07:52 (UTC)

having problems installing with pacaur. It builds but when trying to install it says: error: '/path/to/tmp/netatalk/netatalk-ddp-2.2.2-4-any.pkg.tar.xz': could not find or read package Looks like it expects the ddp-version, even if only the normal is built.

SJ_UnderWater commented on 2012-04-21 04:20 (UTC)

a few more minor fixes in this one, making sure the conflicts go through correctly

SJ_UnderWater commented on 2012-04-10 20:49 (UTC)

sorry about the ddp trouble everyone, turns out it was a fakeroot issue (I often build --asroot). I've still added the "conflicts" flags, but it shouldn't build more than one package ever again. I've also implemented this in a much safer way that should make it resilient with other AUR-building solutions.

carbncl commented on 2012-04-10 20:01 (UTC)

I see, thanks for quick replies, and good job.

SJ_UnderWater commented on 2012-04-10 19:30 (UTC)

the comment about the "first major build" explains the initscript "architecture"

carbncl commented on 2012-04-10 19:25 (UTC)

I see, just tested, works perfectly! Forget about -ddp, Just figured I missed the big picture... What's the purpose of afpd/cnid_metad in /etc/rc.d since netatalk now does the job? (and since I don't think it provides backward compatiblity because of cnid renamed to cnid_metad?)

carbncl commented on 2012-04-10 19:07 (UTC)

I guess you just need to add `conflicts=('netatalk-ddp')` > would be more clean and logical in -ddp PKGBUILD. Just upgraded with your netatalk PKGBUILD, so far no issue. Still using afpd/cnid_metad since I do not want all netatalk services.

SJ_UnderWater commented on 2012-04-10 18:49 (UTC)

i hadn't expected yaourt to build the package. if used "normally" with makepkg, only netatalk will be built by default without any issues, netatalk-ddp will only be built if specified by itself (as in the wiki entry). I'll have to examine yaourt and see what I can do to prevent it from building both at the same time

teek commented on 2012-04-10 12:05 (UTC)

I used Yaourt to install the package put it also gives the below mentioned errors. You say: "leave dpp out" but how do you do that and prevent the: "/usr/share/man/man8/afpd.8.gz exists in both 'netatalk' and 'netatalk-ddp'" Errors? Do I have to edit the PKGBUILD? What should I change exactly?

emphire commented on 2012-04-09 22:23 (UTC)

Thanks SJ. I didn't realize there was a new wiki entry. It was very easy to follow and I'm up and running now.

SJ_UnderWater commented on 2012-04-09 18:06 (UTC)

you're only supposed to install _one_. did you check the wiki entry for netatalk? If you don't need ddp, leave it out.

emphire commented on 2012-04-09 17:57 (UTC)

I get the following errors and lots of similar errors for other files when trying to install the package: Targets (2): netatalk-2.2.2-3 netatalk-ddp-2.2.2-3 Total Installed Size: 6.06 MiB Net Upgrade Size: 1.68 MiB Proceed with installation? [Y/n] (2/2) checking package integrity [############################] 100% (2/2) loading package files [############################] 100% (2/2) checking for file conflicts [############################] 100% error: failed to commit transaction (conflicting files) /etc/netatalk/AppleVolumes.default exists in both 'netatalk' and 'netatalk-ddp' /etc/netatalk/AppleVolumes.system exists in both 'netatalk' and 'netatalk-ddp' /etc/netatalk/afp_ldap.conf exists in both 'netatalk' and 'netatalk-ddp' /etc/netatalk/afpd.conf exists in both 'netatalk' and 'netatalk-ddp' /etc/netatalk/netatalk.conf exists in both 'netatalk' and 'netatalk-ddp' /etc/pam.d/netatalk exists in both 'netatalk' and 'netatalk-ddp' [...]

munkoil commented on 2012-04-08 09:25 (UTC)

Thanks, I'm not familiar with split packages. I saw the missing ' when adding more dependancies to the depends variable but didn't know the resolve_deps command. Now it builds and so far it works like a charm. Will test it further and come back to you if I find anything.

SJ_UnderWater commented on 2012-04-08 08:44 (UTC)

should be fine now, was a split-package incompatibility

munkoil commented on 2012-04-08 06:49 (UTC)

Hi again, Some trouble building and I can't find the problem ==> Entering fakeroot environment... /usr/bin/makepkg: eval: line 1505: syntax error near unexpected token `(' /usr/bin/makepkg: eval: line 1505: `provides_list+=netatalk _options=`echo $_options | sed s/disable/enable/g` build_netatalk } package_netatalk() {' ==> Starting package_netatalk()... -> Building... make: *** No rule to make target `install'. Stop. ==> ERROR: A failure occurred in package_netatalk(). Aborting...

munkoil commented on 2012-04-08 06:27 (UTC)

Hi, Great that you will try to get it accepted into the community repository. When I unzipped the tarball I noticed that all files have execute permission. This is not correct as far as I know. No files need this before they are installed in their final location.

SJ_UnderWater commented on 2012-04-08 03:48 (UTC)

any testing feedback would be greatly appreciated

SJ_UnderWater commented on 2012-04-08 01:44 (UTC)

second build, now uses the release package, git not required. The package has been cleaned up a little more, following the new guidelines. the initscripts now check dependencies in both directions when starting/stopping.

SJ_UnderWater commented on 2012-04-06 05:27 (UTC)

just to clarify, add "netatalk" to you daemons list to launch at startup

SJ_UnderWater commented on 2012-04-04 02:02 (UTC)

ok, the first major build. There are two compilation paths: modern, or modern + crummy old crap. I decided not to support individual features because of the complexity. Both paths compiled for me, let me know of issues. There is now a netatalk initscript that starts the other daemons depending on the features installed, and in a specific order (subject to review). Only the netatalk script registers as a daemon, but doesn't "display". The child scripts display but don't register (you can always run "ps" or check the pids in /var/run). Rc.d doesn't like daemons that launch others, but since the others don't register there's no ambiguity, the launch order is preserved. This way users can also manually stop/restart each service as circumstances dictate. I'm working on completeness, hoping Netatalk will eventually be included in community, so let me know about any edge cases, nitpicks, etc.

carbncl commented on 2012-04-03 21:22 (UTC)

Thanks andrej84 for your contributions on this, appreciated. And welcome to SJ_UnderWater as new maintainer :)

andrej84 commented on 2012-04-03 20:22 (UTC)

I think I needed to abandon it first, now I did and it should be easy to adopt, thanks for that!

andrej84 commented on 2012-04-03 20:12 (UTC)

ok good thanks, if you want you can take over for netatalk anyway, I'm actually not using it anywhere so it's a bit tricky to test for me..

teek commented on 2012-03-06 15:22 (UTC)

On second thought, maybe I'm just seeing the Avahi daemon as a server but the timemachine server does not actually work...

teek commented on 2012-03-06 15:11 (UTC)

Works for me, server appears under network on my mac however when I click it it tells me something like: "Connection failed". By the way, the name is the one from my rc.conf, not from /etc/netatalk/AppleVolumes.default Any suggestions?

karol_007 commented on 2012-01-22 13:00 (UTC)

Works for me, but [karol@black netatalk]$ sudo rc.d start papd :: Starting papd Daemon [DONE] [karol@black netatalk]$ sudo rc.d stop papd :: Stopping papd Daemon [BUSY] rm: cannot remove `/var/run/': No such file or directory [DONE] The daemon starts and stops fine.

andrej84 commented on 2012-01-22 12:38 (UTC)

Well the problem was simply that they disabled appletalk by default, passing --enable-ddp makes it work again. I also moved openssl and libgcrypt to optional dependencies, as they suggest in the official page Please try it out now, if it works we can fix also the netatalk-git packet..

commented on 2012-01-17 21:27 (UTC)

Version 2.2.2 is out now. Hope that version works better

andrej84 commented on 2012-01-14 21:54 (UTC)

Ah then they're both broken (maybe it's worth to notify also to the other one), ok if I can't figure out soon what's happened I will downgrade to the previous versions which I'm sure was actually working..

karol_007 commented on 2012-01-14 21:52 (UTC)

I've tried netatalk-git and got the same results.

andrej84 commented on 2012-01-14 21:51 (UTC)

Actually you're right :O I think that the 2.2.1 is actually broken then, sorry for not noticing. Maybe can you try the netatalk-git package to see if that works?

commented on 2012-01-14 19:34 (UTC)

When i try to start atalkd i get the following error ./atalkd: line 12: /usr/sbin/atalkd: No such file or directory And when i try to start papd i get the following error ./papd: line 10: /usr/sbin/papd: No such file or directory And when I check those directories does not exist Why? What have I done wrong?

munkoil commented on 2011-11-07 13:16 (UTC)

Hi, afpd.service is no longer needed since afpd is now auto registering itself to avahi. By the way, check out netatalk-git to compare the dependencies and so on.

denn commented on 2011-11-01 07:34 (UTC)

pkg-config should be added to makedepends

andrej84 commented on 2011-10-31 22:37 (UTC)

I tried to compile and run and everything seems fine, please test it and let me know if there are problems... Thanks, Andrea

andrej84 commented on 2011-10-25 23:31 (UTC)

Sorry guys, I'm still here yes, in the next days I'll test and get the new version in aur...

emphire commented on 2011-10-15 23:23 (UTC)

Version 2.2.1 is out now. Is the maintainer still around?

carbncl commented on 2011-08-24 15:46 (UTC)

no problem with libfakeroot on my side. maybe that helps might be relate to setuid (I build --asroot, bad me)

andrej84 commented on 2011-08-24 15:12 (UTC)

I've set avahi and tcp_wrappers as optional, removed --without-xfs and added libgcrypt as dependency. Seems to work here... Another thing, do you also get these errors: ERROR: object '' from LD_PRELOAD cannot be preloaded: ignored.

carbncl commented on 2011-08-20 14:31 (UTC)

Since I think v2.1, afpd is auto registering itself to avahi. /etc/avahi/services/afpd.service is not required anymore and can even make afpd fail to start, service being already registered. (issue fixed 2 days ago in 2.2.1) So with that file removed, avahi is definitely optional, but highly recommended. From what I see (and tested), tcp_wrappers can be used by afpd, but is not required. (optional as well) Another thing, libgcrypt should be added as a dependency since it is required for DHX2 authent, a must have for OSX 10.7 Lion. Last point, I guess it disappeared with recent version, but you should remove --without-xfs from configure (unknown option) see netatalk-git PKGBUILD updated with these remarks.

andrej84 commented on 2011-08-20 12:33 (UTC)

@nickl so what should I do? If I'm not able to automatize the application of the patch or - I remove the --enable-fhs - I wait for the next version when it will be fixed. @munkoil Good question, but two files in the package are: netatalk /etc/avahi/services/ netatalk /etc/avahi/services/afpd.service But I'm not sure that means that it's a requirement. Probably tcp_wrappers is not necessary too, any confirmation on this?

munkoil commented on 2011-08-14 06:25 (UTC)

Hi, I've looked over and uninstalled orphans in my system and noticed that there was still one package using tcp_wrappers which was dropped support by arch core library. After looking at the requirements of netatalk I realized that a lot of the packages listed as dependencies above actually is optional and should probably be listed as optdepends. For example I think avahi should actually be optional since it is not required by the package, but recommended in my opinion. What to do with the rest? Can tcp_wrappers be removed or is it needed? Any suggestions?

nickl commented on 2011-08-01 14:45 (UTC)

It's a bug in the new release of netatalk, and will be fixed in the next release: For now, here's a script that will download a patch and builds successfully: It references a patch I put on pastie, not sure how long that will stay up. If you have somewhere better to host it that would be good.

andrej84 commented on 2011-07-28 09:27 (UTC)

I was trying to upgrade, but makepkg fails. While I compile I get this: server_ipc.c: In function ‘reconnect_ipc’: server_ipc.c:195:43: error: ‘_PATH_AFP_IPC’ undeclared (first use in this function) But if I do a clean ./configure && make it works. So I guess that some configure flag is broken, anyone else noticed and can give me a hint?

carbncl commented on 2011-07-27 15:43 (UTC)

@andrej84 stable v2.2 now officially available :

andrej84 commented on 2011-07-27 10:42 (UTC)

Ok I took care of the packet, and I uploaded the new one with all the changes I talked about below. Let me know if there are problems, Andrea

commented on 2011-07-27 10:25 (UTC)

Hey, sorry for any inconvenience, had a lot trouble the last months, and after all I'm not running any arch linux systems. So I gave up the maintaining of the netatalk package.

carbncl commented on 2011-07-24 18:42 (UTC)

That was not that hard after all, netatalk-git package :

carbncl commented on 2011-07-24 18:27 (UTC)

According to HAT, a netatalk contributor, v2.2 or v2.3 is on its way with Lion support : So I have just made a quick (and probably dirty package) with the latest git version, as HAT suggest it will be the next official stable release. Sorry for the lack of packaging, I'm a bit new with arch, but that could be a base for netatalk-git package : 1. Take the the package of x-demon : 2. And just replace with that PKGBUILD:

andrej84 commented on 2011-07-24 13:27 (UTC)

Yes well it would be simpler to upload the package, anyway this is the unified diff for the patch I've applied to update and fix a few issues in the netatalk package.

commented on 2011-07-23 22:43 (UTC)

Package needs update, having problems to access my shares (AFP) from a newly upgraded Mac OS X Lion.

nickl commented on 2011-07-23 12:24 (UTC)

andrej84: Isn't it just a new PKGBUILD? Just post the text of it up to pastebin or something similar. If there are patches that need to be applied, just link or pastebin them too. If you really need to post a tar file put it on megaupload or rapidshare or mediafire.

andrej84 commented on 2011-07-22 14:46 (UTC)

The problem is that I find no service to host a simple tar file (incredibly). Any suggestions about that? Otherwise I'll send it to the maintaner or the mailing list...

nickl commented on 2011-07-22 11:56 (UTC)

andrej84. yes! and also post it up on pastebin or the like and drop a link here. if wodar doesn't respond, you can email the aur list and ask them to orphan the package and you can become a maintainer.

andrej84 commented on 2011-07-22 09:16 (UTC)

I created a new package for 2.1.15 with the fixes of all the issues cited here (PID quoted, cdb support and 744 permissions to afpd.service). Should I send it to the responsible for netatalk?

munkoil commented on 2011-07-10 06:35 (UTC)

I'm having trouble with netatalk after upgrading db to 5.2.28-1. It works fine when I downgrade to 5.1.25-1 again. Any suggestions?

commented on 2011-05-07 14:54 (UTC)

i made quick and dirty fix - at least with that version i can use timemachine in mac os x 10.7 and it does not throws shitload of error anymore.

commented on 2011-04-20 00:49 (UTC)

It would be nice to see the package updated to 2.2 beta or at least 2.1.5 THANKS

emphire commented on 2011-03-26 00:14 (UTC)

Note that pkgrel usually starts at 1, not 0. Thanks!

commented on 2011-03-13 02:34 (UTC)

I had to put quotes around $PID on lines 10 and 22 of /etc/rc.d/afpd, otherwise bash complained about a binary operator expected.

link commented on 2011-02-01 17:07 (UTC)

The installed afpd.service is installed with permissions of 755. I don't really see why a config file should be installed with the executable bit set.

commented on 2010-12-17 11:34 (UTC)

@all would integrate all your suggestions into the next minor release from netatalk. Where already the fourth alpha version is out. If you want the feature now, please tell me, i will then push out a update with just the scripts and the added configure parameter. @frank_s could you please explain what kind of error you had experienced. My both servers are running just fine with dbd and the latest Mac OS X 10.6.5 Client.

commented on 2010-12-11 20:58 (UTC)

Please add to configure section "--with-cnid-cdb-backend" because with latest MacOS X 10.6 an error occur with the dbd backend. With cdb backend netatalk work as expected.

commented on 2010-12-08 15:38 (UTC)

It looks like the init script script has problems starting/stopping afpd if multiple afpd processes are spawned. This seems to be normal behaviour for afpd (?). If you would like to look at my patched init script let me know.

commented on 2010-11-03 12:32 (UTC)

with dropping the db4.1 dependency, you could run into something like removing the old version and install a new version of db. This will probably get you to cnid errors. More precise cnid_dbd will say it can not load the libdb-4.8 lib. The quick solution to fix this is just to reinstall the package, ite will compile with the new db version and everything should be back to normal.

commented on 2010-10-29 23:29 (UTC)

- update to current minor version - changed dependency from db4.1 after suggestions from richjn and anders - added a afpd.service like advised from munkoil thanks for the feedback and happy talkn

commented on 2010-10-29 14:24 (UTC)

please drop the db4.1 dependency

richjn commented on 2010-10-04 07:27 (UTC)

Hi Thanks for the package. I beleive you can drop the DB4.1 dependency now as it will compile / run fine with DB5 (fixed in version 2.1.2) Richard

munkoil commented on 2010-09-24 20:45 (UTC)

I've been thinking about if a configuration file for avahi should be included in the package since avahi is a requirement. I don't know if it is the correct thing to do, it's just a suggestion. The following file will make the afp server show up in finder automatically if avahi-daemon is started /etc/avahi/services/afpd.service <?xml version="1.0" standalone='no'?> <!DOCTYPE service-group SYSTEM "avahi-service.dtd"> <service-group> <name replace-wildcards="yes">%h</name> <service> <type>_afpovertcp._tcp</type> <port>548</port> </service> <service> <type>_device-info._tcp</type> <port>0</port> <txt-record>model=Xserve</txt-record> </service> </service-group>

commented on 2010-08-23 18:58 (UTC)

Currently stoping the afpd daemon with connected clients will fail! You can force it with killall. Searching for a "nice" way to clean up and not to be too rude to the clients. Any suggestions?

commented on 2010-08-23 18:38 (UTC)

- new version - correct cnid start script - correct afpd start script think the atalkd and papd need some love.

risk commented on 2010-07-20 05:57 (UTC)

sudo /usr/sbin/cnid_meta will make it daemonize and afpd with dbd will just magically work.

munkoil commented on 2010-07-13 17:41 (UTC)

Hi, Thanks for the PKGBUILD! The package is outdated. When you update the package can you please fix afpd so it reads from netatalk.conf. Currently my CNID_METAD is not started and I have not been able to sort it out. If anyone know how to fix this please feel free to post a comment.

farhany commented on 2010-07-02 04:11 (UTC)

I'll take a look at that soon.

commented on 2010-07-02 04:04 (UTC)

Hi. First, thanks for the PKGBUILD. This saved me a chunk of time :) The main issue I had was that the netatalk.conf file is not adhered to at all by the rc.d files. In particular, cnid_metad is not started anywhere.