gvfsd-ftp - Pure-FTPd Error: Invalid reply

Thank you for your bug report. Are you sure this ftp is working in anonymous? Neither gftp nor lftp is able to connect to it either

No no, it requires ftp login and password, I just gave an example of a server that uses pure-ftpd.
Try ftp from the terminal

I mean compare:
nautilus ftp://eclipse.sh3lls.net/
and
ftp eclipse.sh3lls.net

nautilus pops up an error, ftp doesn't.

$ lftp ftp://eclipse.sh3lls.net
lftp eclipse.sh3lls.net:~> debug 3
lftp eclipse.sh3lls.net:~> ls
---- Connecting to eclipse.sh3lls.net (66.252.23.2) port 21
<--- 220---------- Welcome to Pure-FTPd [TLS] ----------
<--- 220-You are user number 1 of 50 allowed.
<--- 220-Local time is now 16:50. Server port: 21.
<--- 220-This is a private system - No anonymous login

the ftp doesn't accept anonymous logins

I was about to cross the line here, but now I understand why you're emphasizing on this.
I am not asked for a login info, the error is shown without asking me if I want to login.

$ lftp ftp://<email address hidden>
Password:
lftp <email address hidden>:~> debug 3
lftp <email address hidden>:~> ls
---- Connecting to eclipse.sh3lls.net (66.252.28.2) port 21
<--- 220---------- Welcome to Pure-FTPd [privsep] [TLS] ----------
<--- 220-You are user number 1 of 50 allowed.
<--- 220-Local time is now 18:25. Server port: 21.
<--- 220-This is a private system - No anonymous login
<--- 220-IPv6 connections are also welcome on this server.
<--- 220 You will be disconnected after 15 minutes of inactivity.
<--- 230-User someuser has group access to: someuser
<--- 230 OK. Current restricted directory is /
drwx--x--x 6 someuser someuser 512 Mar 17 23:23 .
drwx--x--x 6 someuser someuser 512 Mar 17 23:23 ..
-rw------- 1 someuser someuser 6330 Mar 25 14:34 somefile

what version of gvfs do you use? could you run /usr/lib/gvfs -r on a command line and try to connect and note what happens? The authentification dialog is correctly displayed on my hardy installation

That's odd, I tried another server, to my knowledge that uses pure-ftpd, ftp://ftp.greek-fun.com - this one asks me for password.

I have tried what you said on eclipse.sh3lls.net:
$ /usr/lib/gvfs -r
bash: /usr/lib/gvfs: is a directory

$ /usr/lib/gvfs/gvfsd -r
Added new job source 0x642010 (GVfsBackendFtp)
Queued new job 0x643030 (GVfsJobMount)
<-- 220---------- Welcome to Pure-FTPd [privsep] <-- [TLS] ----------
send_reply, failed: 1

Version:
gvfs-backends: 0.2.2-1
gvfs: 0.2.2-1

It looks like I have the same thing on all the servers on *.sh3lls.net, some that I tried: vortex astro dasher eclipse

not confirming, that works correctly on my hardy installation, to send upstream by somebody having the issue

Ok let me give some more info:
I have been using hardy since alpha 5, I kept /home partition intact, and formatted the root / partition
I'm using x86_64 / amd64 ubuntu

I've tried to compare the outputs of /usr/lib/gvfs/gvfsd -r:
1) nautilus ftp://<email address hidden>
Added new job source 0x642010 (GVfsBackendFtp)
Queued new job 0x643030 (GVfsJobMount)
<-- 220---------- Welcome to Pure-FTPd [privsep] <-- [TLS] ----------
send_reply, failed: 1

2) nautilus ftp://<email address hidden>
Added new job source 0x642010 (GVfsBackendFtp)
Queued new job 0x643030 (GVfsJobMount)
<-- 220---------- Welcome to Pure-FTPd [TLS] ----<-- ------
send_reply, failed: 1

3) nautilus ftp://<email address hidden>
Added new job source 0x642010 (GVfsBackendFtp)
Queued new job 0x643030 (GVfsJobMount)
<-- 220---------- Welcome to Pure-FTPd [TLS] ----<-- ------
send_reply, failed: 1

4) (the working one) nautilus ftp://<email address hidden>
Added new job source 0x642010 (GVfsBackendFtp)
Queued new job 0x643030 (GVfsJobMount)
<-- 220---------- Welcome to Pure-FTPd [TLS] ----------
<-- 220-You are user number 1 of 50 allowed.
<-- 220-Local time is now 20:22. Server port: 21.
<-- 220-IPv6 connections are also welcome on this server.
<-- 220 You will be disconnected after 15 minutes of inactivity.
(I hit cancel at the login prompt)
send_reply, failed: 1

Now I've compared it in telnet:
$ telnet eclipse.sh3lls.net 21
Trying 66.252.23.2...
Connected to eclipse.sh3lls.net.
Escape character is '^]'.
220---------- Welcome to Pure-FTPd [TLS] ----------
220-You are user number 1 of 50 allowed.
220-Local time is now 17:26. Server port: 21.
220-This is a private system - No anonymous login
220-IPv6 connections are also welcome on this server.
220 You will be disconnected after 15 minutes of inactivity.

$telnet dasher.sh3lls.net 21
Trying 66.252.13.2...
Connected to dasher.sh3lls.net.
Escape character is '^]'.
220---------- Welcome to Pure-FTPd [privsep] [TLS] ----------
220-You are user number 1 of 50 allowed.
220-Local time is now 12:26. Server port: 21.
220-This is a private system - No anonymous login
220-IPv6 connections are also welcome on this server.
220 You will be disconnected after 15 minutes of inactivity.

$ telnet astro.sh3lls.net 21
Trying 69.42.213.130...
Connected to astro.sh3lls.net.
Escape character is '^]'.
220---------- Welcome to Pure-FTPd [TLS] ----------
220-You are user number 1 of 50 allowed.
220-Local time is now 17:27. Server port: 21.
220-This is a private system - No anonymous login
220-IPv6 connections are also welcome on this server.
220 You will be disconnected after 15 minutes of inactivity.

See the difference of the last "----<-- ------" or "[privsep] <-- [TLS] ----------"
Looks like their server is broken? Or something in gvfsd parses the banner wrongly?

I've used aptitude purge to completely remove the packages gvfs gvfs-backends libgvfscommon0 nautilus nautilus-data, removed the downloaded copies from /var/cache/apt/archives/ and redownloaded and installed ubuntu-desktop gvfs-backends libgvfscommon0 nautilus nautilus-data (and all the other stuff that came with it). Nothing changed.

I too have this error.

I have a working ftp server that does not accept anonymous connections.

My system is fully up-to-date with with all the updates @ 22:00 GMT

My server I connect to is ProFTPD 1.3.1 Server running on FreeBSD.

Only nautilus gives the error Invalid reply.

Is there some debugging I can run to check?

Hm.. the sh3lls servers are FreeBSD too. think there's a connection? But as Sebastien mentioned, he didn't get that error, that's what makes it weird.

Interestingly enough, I installed proftpd locally and it works fine.
First is the Ubuntu server, then the BSD one

I used the same daemon version for comparison purposes. No idea what the difference in response is apart from the custom server name.

Still the Ubuntu one works, the BSD one does not.

Connected to localhost.
220 ProFTPD 1.3.1 Server (Debian) [127.0.0.1]
Name (localhost:xxx): xxx
331 Password required for xxx
Password:
230 User xxx logged in
Remote system type is UNIX.
Using binary mode to transfer files.
ftp> bye
221 Goodbye.

Connected to tokra
220 ProFTPD 1.3.1 Server (BSD) [192.168.0.6]
Name (tokra:xxx): xxx
331 Password required for xxx
Password:
230 User xxx logged in
Remote system type is UNIX.
Using binary mode to transfer files.
ftp>

Forgot to add:

If I use Firefox to connect to the BSD server, it all works perfectly to the remote server

-G

Try the steps I did:
/usr/lib/gvfs/gvfsd -r
Then open the connection in a new terminal:
nautilus ftp://server.com

...the gvfsd -r should show an output of what's going on

> Try the steps I did:
>
> /usr/lib/gvfs/gvfsd -r
> Then open the connection in a new terminal:
> nautilus ftp://server.com
>

you guys might want to comment upstream rather, they write the code ubuntu only distribute it, they might have a better idea on the issue

OK, I found out something rather interesting.

I saw an anomaly using the -r switch.

IF the server's description text is more than 3 characters, then the reply would fail.

It would appear that the buffer is too small to hold this string.

For verbosity, I have made the reply name 1234567890, then reducing each name as we go on

Here are the results:

<-- <-- 220 ProFTPD 1.3.1 Server (1234567890) [192.16<-- 8.0.6]
send_reply(0x80868b8), failed=1 (Invalid reply)
no more connections left, exiting...Added new job source 0x807a800 (GVfsBackendFtp)
Queued new job 0x807bc18 (GVfsJobMount)
<-- 220 ProFTPD 1.3.1 Server (123456789) [192.168<-- .0.6]
send_reply, failed: 1

Added new job source 0x807a800 (GVfsBackendFtp)
Queued new job 0x807bc18 (GVfsJobMount)
<-- 220 ProFTPD 1.3.1 Server (12345678) [192.168.<-- 0.6]
send_reply, failed: 1

Added new job source 0x807a800 (GVfsBackendFtp)
Queued new job 0x807bc18 (GVfsJobMount)
<-- 220 ProFTPD 1.3.1 Server (1234567) [192.168.0<-- .6]
send_reply, failed: 1

Added new job source 0x807a800 (GVfsBackendFtp)
Queued new job 0x807bc18 (GVfsJobMount)
<-- 220 ProFTPD 1.3.1 Server (123456) [192.168.0.<-- 6]
send_reply, failed: 1

Added new job source 0x807a800 (GVfsBackendFtp)
Queued new job 0x807bc18 (GVfsJobMount)
<-- 220 ProFTPD 1.3.1 Server (12345) [192.168.0.6<-- ]
send_reply, failed: 1

Added new job source 0x807a800 (GVfsBackendFtp)
Queued new job 0x807bc18 (GVfsJobMount)
<-- 220 ProFTPD 1.3.1 Server (1234) [192.168.0.6]<--
send_reply, failed: 1

Added new job source 0x807a800 (GVfsBackendFtp)
Queued new job 0x807bc18 (GVfsJobMount)
<-- 220 ProFTPD 1.3.1 Server (123) [192.168.0.6]
--> USER xxx
<-- 331 Password required for xxx

I must also add in that '/usr/lib/gvfs/gvfsd -r' appears only to be doing something when the connection is to a remote server. Somehow local connections do not create an entry..

confirmation report that I seem to have this same bug connecting to my web server via nautilus; command line and firefox work, but nautilus doesn't (I get the invalid reply message). This only happened on Intrepid, on Hardy and previous I didn't have this problem. (my provider is also using pure-ftpd, which is why I think it might be related).
If there are any tests/outputs you would like, let me know what they are and I will post here.

I have the same problem. It only happens on one of my Nautilus bookmarked FTP sites, others are fine. I have three computers running Hardy on two different networks and it only happens on one of them. Here is the response I get.

  ~$ /usr/lib/gvfs/gvfsd -r
Added new job source 0x807a800 (GVfsBackendFtp)
Queued new job 0x807bc18 (GVfsJobMount)
<-- 220 FTP Server ready.<--
send_reply, failed: 1

So? no fix yet?

Daniel: a patch exists in http://bugzilla.gnome.org/show_bug.cgi?id=525283 but noone with an account on the broken servers has yet offered to test it.

could somebody having the issue try the upstream change?

Andreas' fix committed upstream!!

        reviewed by: Andreas Henriksson <email address hidden>

        Bug 525283 - handle short reads in ftp

        * daemon/gvfsbackendftp.c: (ftp_connection_receive):
        account for cases where soup_socket_read_until() would not read up to
        the boundary on the first read.

This bug was fixed in the package gvfs - 1.1.7-0ubuntu1

---------------
gvfs (1.1.7-0ubuntu1) jaunty; urgency=low

  * New upstream version:
    - Fix build on some platforms
    - ftp: Fix short read errors (lp: #208750)
    - gphoto2: Make it work on iphone
    - sftp: Fix symlink creation
    - fuse: Better support of truncation, fixing OOo save (lp: #317587)
    - proxy monitors: Support mount operations, etc
    - gvfs-mount: add --device commandline support
    - sftp: Fix protocol bug that made some servers not work (lp: #286053)
    - general support for query_info over streams, implemented for smb, sftp
    - Initial support for .xdg-volume-info reading (lp: #269159)
    - fix trash crasher (lp: #332554)
    - use ssh config (lp: #264803)
  * debian/patches/02_support_xdg_volume_info.patch:
    - the new version fixes this issue
  * debian/patches/91_upstream_change_fix_trash_crasher.patch:
    - upstream change to fix a trash crasher (lp: #333791)
  * debian/rules:
    - shlibs version update

 -- Sebastien Bacher <email address hidden> Mon, 02 Mar 2009 18:10:40 +0100