Activity for Daniel Markstedt
-
Daniel Markstedt modified ticket #513
afpd panic on NexentaOS (3.0.3)
-
NexentaOS is long discontinued at this point. We are continuously testing on Solaris 11 and OmniOS: dtrace works in 3.2.5 and the main development branch. So this was either specific to NexentaOS, or since fixed.
-
Allow installation for multiple init system
-
Resolved with https://github.com/Netatalk/netatalk/pull/1371 and will be in the future 4.0.0 release.
-
netatalk-3.11.1 failed on CentOS7.3.1611 at afpd load
-
This bug has been obsoleted by the Meson build system.
-
afpd crash (netatalk 3.1.11) on OmniOS (pkgsrc)
-
Appletalk write lock isn't propogated to SMB
-
Much improved in netatalk3. See https://github.com/Netatalk/netatalk/wiki/FAQ#user-content-How_does_Netatalk_integrate_with_Samba
-
Slow performance when opening files with specific applications (eg. mkv with vlc)
-
I've played mkv files over an AFP share recently without this problem.
-
Network trash not always available under Lion
-
Affecting long-obsoleted versions of macOS
-
The operation can’t be completed because the item “ ” is in use
-
Reading the Apple support thread, it seems like the cnid database needs rebuilding. AppleDouble metadata might have gone out of sync when messing directly with the file system. If on OSX/macOS, avoid using AppleDouble, rather use EA with netatalk3.
-
netatalk time machine backup could not complete
-
TM is generally known to work with recent netatalk & macOS versions
-
Quotation marks in afpd.conf break path resolution
-
TM hangs endless and stops keyboard/harddisk access
-
The baseline TM setup works with the latest netatalk and macOS
-
This would be helpful for Debian as well, since systemd is preferred but init.d rc scripts is recommended as a fallback for packaged software. (lintian complains when fallback is missing)
-
Very slow TimeMachine Backups from Yosemite
-
dbus-daemon: dbus Rejected send message, 2 matched rules
-
AFP_DELETE -> AFPERR_ACCESS
-
The appledouble parsing and validation code has changed recently.
-
afp issue with InDesign CC 2015 /Quark not packaging links
-
version of the server you are trying to connect to is not supported
-
Media stream crashes
-
not replace variables in documents
-
Bug migrated to github for tracking https://github.com/Netatalk/netatalk/issues/383
-
Thanks for responding. I'm going through the bug backlog to try to assess which ones are still relevant. Netatalk saw little development for the last decade after Apple deprecated AFP. So most bugs as old as this are still valid.
-
So if I understand the user scenario correctly, this is an iTunes music library that is stored on an AFP volume, right?
-
How slow is "very slow" and what kind of performance are you expecting? I've been doing TM backups with my Ventura system for a few weeks, and I agree it isn't super fast, but I'm not sure exactly what a good baseline is.
-
The first thing to confirm here is that both dbus config files are in the right places and readable by dbus/netatalk. For one, do "netatalk -V" and then confirm that dbus-session.conf is in the right path as indicated there. And secondly, make sure "netatalk-dbus.conf" is the dbus systemd config dir, e.g. /etc/dbus-1/system.d
-
Can you please clarify something: By "apple time capsule", do you mean an actual time capsule device's file system is mounted remotely and then shared by netatalk? Or is this a plain netatalk volume of the time machine type? Looking at the log, the one hint that I see is that EA metadata for the file cannot be read, so the format of the metadata might be different than what netatalk expects, or it has been corrupted somehow.
-
Can you please provide more details? What is the "configuration fce notify script"?
-
I tested this with the latest netatalk development code in the main branch, on a macOS Ventura client, streaming mkv files from an afp shared volume, using the VideoLAN client. This worked alright, although I got pauses for buffering, probably due to latency over wifi. In general I don't think AFP is an ideal protocol to stream video. This is not what it was designed for. To answer your questions: I do believe that a video would be read by a media player client as a stream over AFP. I do not think...
-
For the former log, this is encapsulated in a HAVE_NFSV4_ACLS check so if you don't actually need ACL support you can suppress this one by configuring netatalk with "--with-acls=no", build and install. For the latter log, as far as I can tell from the code, it is a debug level log, so it should be muted if you set log level to "info" in afp.conf, for instance. BTW the FreeNAS bug tracker is long gone, but the bug ticket is archived at https://web.archive.org/web/20150928131655/https://bugs.freen...
-
Was this issue specific to when compiling from the srpm? I set up a CentOS Stream 9 environment to test this and was able to get afpd to load uams_gss.so without error, using the latest netatalk main branch development code. The resulting config.h looked like the tarball example that HAT provided.
-
File system freeze after "pkill -HUP afpd"
-
Would you be able to provide logs leading up to the error? And to confirm, you're saying that the entire Centos system runs on a StorNext filesystem? Or is it just the shared netatalk volume that is? My hunch is that the filesystem is a factor here, so please help me understand how you have configured it. And if possible, I'd like to try and replicate your setup, so any pointers on how to set up StorNext with Centos would be appreciated!
-
What are the typical steps leading up to the crash? The logs talks about pdf and psd files. How are these files created? Did they exist at the time of the crash? FWIW I've been running the latest netatalk main development code on FreeBSD13 for the last week without incident, but I haven't done any really advanced file operations. Perhaps it is a particular app that generates files with particular metadata that causes this issue?
-
I've tried to get netatalk3 to compile on OmniOS r151044 but am really struggling to get a working libgcrypt set up. Do you have any pointers on how to prepare a full compilation environment?
-
The two reports don't seem to have much in common apart from both using TM volumes and both specifying the "valid users" parameter for the TM volume. Just as a troubleshooting step, does it make a difference if you remove that parameter? We've had known corner case scenarios with users and permissions over the years... FWIW, I've been running a TM server with the latest main development code on Debian, and my Ventura MBP over the last 2 weeks with hourly backups and not had major issues, so it should...
-
So if I get this correctly, the actual dir structure on the system remains as /mnt/dsl/home/[username] but netatalk looks in /mnt/dsl/home/AUTH/[username]? I wonder if the update made it so that netatalk reads from an afp.conf file in a different location than usual, which is misconfigured with "basedir regex = /mnt/dsl/home/AUTH". Can you please run "netatalk -V" on the command line and check where it's reading afp.conf from?
-
FWIW I don't think it's a good practice to have both a dir and a subdir to that dir defined as shared volumes. I think they should be kept nested separately. Do you have a real-life usecase where you actually want to nest them like this? That said, a more deterministic behavior on netatalk's behalf is definitely preferable.
-
The TLD looks incorrect there, with the trailing period in "local." Can you please try without it?
-
Can you please share your afp.conf? I'm also curious if your system supports sendfile() system calls or not. What does the output from configure say about that? FWIW I've been running the latest main development code on Debian Linux with hourly TM backups for my macOS 13 machine for two weeks without major incident.
-
Time Machine Periodically Fails due to authentication issue
-
Reading the freenas bug report, the root cause is pretty clear: Zeroconf registers the AFP server as ... afp://user@host._afpovertcp._tcp.local/Time Machine whereas the macOS keychain stores the credentials for host.local without the afpovertcp piece. (The bug tracker is offline so the report can be found here: https://web.archive.org/web/20151029045756/https://bugs.freenas.org/issues/4808 ) This string is defined in https://github.com/Netatalk/netatalk/blob/main/etc/netatalk/afp_zeroconf.h #define...
-
Time Machine backup unreliable
-
Tentatively closing this. The reports are against ancient versions of netatalk. I've been running hourly backups for two days with 3.1.15 pre-release code, and a Ventura Mac, so far without incident. I will keep this running for at least a month to see how reliable it is.
-
TimeMachine volume fails, is a reserved name?
-
I think this has been resolved long ago. With 3.1.15 prerelease code, I can define TM volumes with spaces in them, and having a macOS Ventura system back up hourly for the last day or so.
-
Thanks for providing the additional context! Yes randnum is a special case because it uses afppasswd for managing credentials in /usr/local/etc/afppasswd (by default) so it is not suprising that you saw different behaviors there. That exact dsi_stream_read error is very very common still today and is because of messy handling of timeouts I think. FYI in the upcoming netatalk release we introduce a cleaner way to share home dirs with specific users or groups, f.e.: [Homes] basedir regex = /home valid...
-
FPDisconnectOldSession won't work
-
Merged.
-
Make Fails on Linux Mint 19-3
-
afpstats fails with dbus exception "Could not get owner of name 'org.netatalk.AFPStats'"
-
Multiple users only one user get access
-
Netatalk 3.0 "valid users" does not work on Homes
-
Merged.
-
Lots of lines with "check_name: illegal name" in syslog
-
Merged.
-
`guest account` man page entry incorrectly requires quoting.
-
Merged.
-
A security hole on afp.conf, when MySQL is CNID backend
-
Merged.
-
Would it make a difference to bump the value used in this macro? (dictionary.c) /** Maximum value size for integers and doubles. */ #define MAXVALSZ 1024
-
Actually, so for testing I figured that I can just validate that afpd is able to read a read-protected afp.conf with this patch. And yes: before the patch, a non-privileged user is refused connections, while with the patch they are able to connect. Raised a PR https://github.com/Netatalk/netatalk/pull/300
-
By following Ralph's advice, the complete patch would look like this: diff --git a/libatalk/util/netatalk_conf.c b/libatalk/util/netatalk_conf.c index d2d16ca9..43306065 100644 --- a/libatalk/util/netatalk_conf.c +++ b/libatalk/util/netatalk_conf.c @@ -1624,7 +1624,9 @@ int load_volumes(AFPObj *obj, lv_flags_t flags) /* try putting a read lock on the volume file twice, sleep 1 second if first attempt fails */ + become_root(); fd = open(obj->options.configfile, O_RDONLY); + unbecome_root(); while...
-
A security hole on afp.conf, when MySQL is CNID backend
-
Is this patch still required? I don't have a mysql setup to test against right now.
-
Filed as PR https://github.com/Netatalk/netatalk/pull/299
-
Netatalk 3.0 admin group signal fault 11
-
Netatalk 3.0 admin group signal fault 11
-
This does not reproduce in post 3.1.14 development code. The manpage docs have been updated since, too.
-
Netatalk 3.0 admin group signal fault 11
-
Netatalk 3.0 "valid users" does not work on Homes
-
A security hall on afp.conf, when MySQL is CNID backend
-
Unexpected disconnection
-
Please try using netatalk3 which has better support for Time Machine.
-
Serve empty EA data when none exist on server
-
Error Compiling version 3.0.1
-
It seems you're missing the openssl libraries, e.g. libssl-dev on Debian
-
FWIW quotation mark parsing was deliberately removed from afp.conf in https://github.com/Netatalk/netatalk/commit/3343c2
-
Lion cannot find Time Machine disk on 3.0.1.4
-
afpstats don't show MacOS 9
-
Works for me: Apr 10 17:03:07 rasp32 afpd[29485]: Login by dmark (AFP2.2) dmark@rasp32:~/dev/netatalk $ afpstats name: dmark, pid: 29485, logintime: Apr 10 17:03:07, state: active, volumes: dmark name: pucko-2, pid: 29420, logintime: Apr 10 16:38:23, state: sleeping, volumes: pucko-2
-
Mac freeze when save with netatalk 3.0.4
-
There is not enough context to provide informed advice. Please raise a new issue ticket in the github project if this issue persists.
-
stack trace when accessing folder that is a symbolic link
-
As per the linked ticket, this was subsequently fixed in 3.1.x.
-
Panic on copying/moving file within one share when destination filesystem differs (bind mount)
-
Enabling Time Machine (3.1) results in Crash
-
iTunes fails to launch after upgrade from 3.0.4 to 3.1.0
-
Closing as tentative duplicate with https://sourceforge.net/p/netatalk/bugs/589/
-
Move files between ZFS datasets not working in 3.0.5
-
Cannot set Finder tags (10.9)
1 >