Issue Proftpd won't start

[Pandy_Legend]

Earlier this week, I upgraded from Plesk 12.5 to Onyx. The upgrade was successful and I encountered no issues at the time, even after a reboot of the server.

Yesterday, I rebooted the server again after adding and configuring a new hard disk. No other changes were made to configuration.

Now I cannot connect to the server via FTP.

From the command line, when running 'proftpd' initially I got the error:

unable to use '/var/run/proftpd/scoreboard': No such file or directory

I followed instructions in https://kb.plesk.com/en/127535 but now I get the following error:

fatal: unable to get connection info: Socket operation on non-socket

It then suggests I user 'ServerType standalone' in the config file. I tried this, but although proftpd successfully starts, any time I try to connect, it immediately closes the connection 'gracefully'.

The symptoms seem very similar to the thread at https://talk.plesk.com/threads/plesk-12-5-rtm-cannot-connect-to-ftp.334726/ so I tried:

rpm -e --nodeps psa-proftpd
plesk installer --select-release-current --install-component=proftpd

But, unlike for the OP of that thread, this has made no difference.

I have tried restarting xinetd, psa and the server itself to no avail. All other services appear to be running fine.

The server is running CentOS 7 (64 bit) with Plesk Onyx. All current updates installed.

 

[IgorG]

What sort of error message do you have after reinstallation this component with

# plesk installer --select-release-current --remove-component=proftpd
# plesk installer --select-release-current --install-component=proftpd

and start proftpd service in command line?

 

[Pandy_Legend]

Hi Igor- thanks for the reply.
When I run the first command, I get the following:

ERROR: To remove some of the requested components, you also need to explicitly remove the following components: Plesk (panel)
Not all packages were installed.
Please resolve this issue and try to install the packages again.
If you cannot resolve the issue on your own, contact product technical support for assistance.
exit status 1

Is what you suggested not effectively the same as:

rpm -e --nodeps psa-proftpd
plesk installer --select-release-current --install-component=proftpd

though?

 

[IgorG]

Well, I see following on my CentOS7 with Onyx:

# rpm -e --nodeps psa-proftpd
# plesk installer --select-release-current --install-component=proftpd
# systemctl restart xinetd
# lsof -i tcp:21
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
xinetd 24557 root 5u IPv6 23820695 0t0 TCP *:ftp (LISTEN)

Do you have the same result?

 

[Pandy_Legend]

COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
xinetd 39051 root 5u IPv6 1317636 0t0 TCP *:ftp (LISTEN)

 

[IgorG]

So, ftp port 21 is up and listening. What is wrong?

 

[Pandy_Legend]

As stated in my original post, proftpd won't start! It was never a port issue.

2016-12-07 09:50:48,705 hosrvweb01.packimpex.ch proftpd[39618]: processing configuration directory '/etc/proftpd.d'
2016-12-07 09:50:48,705 my.domain.ch proftpd[39618] 127.0.0.1: 192.168.70.10:21 masquerading as 185.67.196.228
2016-12-07 09:50:48,707 my.domain.ch proftpd[39618] 127.0.0.1: fatal: unable to get connection info: Socket operation on non-socket
2016-12-07 09:50:48,707 my.domain.ch proftpd[39618] 127.0.0.1: (Running from command line? Use `ServerType standalone' in config file!)

 

[IgorG]

What is output of command:

# telnet localhost 21

?

 

[Pandy_Legend]

Trying 127.0.0.1...
Connected to localhost.
Escape character is '^]'.
220 ProFTPD 1.3.6rc2 Server (ProFTPD) [127.0.0.1]

Upon looking again, it appears that ProFTPD is running, but not rejecting connections.

07.12.2016 11:00:09 Recv> 220 ProFTPD 1.3.6rc2 Server (ProFTPD) [185.67.196.228]
07.12.2016 11:00:09 Sent> USER xxx
07.12.2016 11:00:09 Recv> 331 Password required for xxx
07.12.2016 11:00:09 Sent> PASS ********
07.12.2016 11:00:09 Recv> 230 User xxx logged in
07.12.2016 11:00:09 Sent> FEAT
07.12.2016 11:00:09 Recv> 211-Features:
07.12.2016 11:00:09 AUTH TLS
07.12.2016 11:00:09 Recv> CCC
07.12.2016 11:00:09 Recv> EPRT
07.12.2016 11:00:09 EPSV
07.12.2016 11:00:09 HOST
07.12.2016 11:00:09 LANG en-US.UTF-8*
07.12.2016 11:00:09 MDTM
07.12.2016 11:00:09 MFF modify;UNIX.group;UNIX.mode;
07.12.2016 11:00:09 MFMT
07.12.2016 11:00:09 Recv> MLST modify*;perm*;size*;type*;unique*;UNIX.group*;UNIX.mode*;UNIX.owner*;
07.12.2016 11:00:09 PBSZ
07.12.2016 11:00:09 PROT
07.12.2016 11:00:09 REST STREAM
07.12.2016 11:00:09 SIZE
07.12.2016 11:00:09 SSCN
07.12.2016 11:00:09 TVFS
07.12.2016 11:00:09 UTF8
07.12.2016 11:00:09 211 End
07.12.2016 11:00:09 Sent> OPTS UTF8 ON
07.12.2016 11:00:09 Recv> 200 UTF8 set to on
07.12.2016 11:00:09 Sent> TYPE A
07.12.2016 11:00:09 Recv> 200 Type set to A
07.12.2016 11:00:09 Sent> SYST
07.12.2016 11:00:09 Recv> 215 UNIX Type: L8
07.12.2016 11:00:09 Sent> TYPE A
07.12.2016 11:00:09 Recv> 200 Type set to A
07.12.2016 11:00:09 Sent> PWD
07.12.2016 11:00:09 Recv> 257 "/" is the current directory
07.12.2016 11:00:09 Sent> CWD /portal-dev
07.12.2016 11:00:09 Recv> 250 CWD command successful
07.12.2016 11:00:09 Sent> PASV
07.12.2016 11:00:09 Stat> Disconnected.
07.12.2016 11:00:09 Unable to load ftp://[email protected]/portal-dev: Socket Error # 10054
Connection reset by peer.

I changed it to stop using passive mode and now it connects successfully.

Whilst this is, of course, a happy resolution it does concern me that this can apparently change without an obvious reason.

I'd be interested to know

1) What could have caused the initial problem of the missing scoreboard file? As this is a known issue since at least the last 2 versions of Plesk, have any steps been taken to overcome it?
2) Why is passive mode now suddenly a problem? Usually I would expect passive mode to be the more reliable of the two.
3) What is the cause of the "fatal: unable to get connection info: Socket operation on non-socket" error and should I be concerned about it?