NTP client hangs in "reached" state

I am trying to synchronize Mikrotik RB750 with a NTP server running on Windows XP. But Mikrotik’s NTP client stuck in reached state and does not want to move forward to timeset or synchronized states (as stated in NTP Server and Client chapter of documentation). Local clock does not change its value. NTP package v4.11 is installed and enabled. NTP server configuration is:

[admin@MikroTik] > /system ntp client print 
        enabled: yes
           mode: unicast
    primary-ntp: 192.168.0.1
  secondary-ntp: 0.0.0.0
         status: reached

In firewall rules I created rule to log packets and I see that responses from NTP server are reaching RB750’s NTP client (192.168.0.1 is WinXP workstation 192.168.0.30 is a RB750):

13:17:29 system,info NTP params changed by admin 
13:17:35 firewall,info input: in:br1 out:(none), src-mac xx:xx:xx:xx:xx:xx, proto UDP, 192.168.0.1:123->192.168.0.30:123, len 56 
13:17:51 firewall,info input: in:br1 out:(none), src-mac xx:xx:xx:xx:xx:xx, proto UDP, 192.168.0.1:123->192.168.0.30:123, len 56 
13:17:53 firewall,info input: in:br1 out:(none), src-mac xx:xx:xx:xx:xx:xx, proto UDP, 192.168.0.1:123->192.168.0.30:123, len 56 
13:17:54 firewall,info input: in:br1 out:(none), src-mac xx:xx:xx:xx:xx:xx, proto UDP, 192.168.0.1:123->192.168.0.30:123, len 56 
13:17:55 firewall,info input: in:br1 out:(none), src-mac xx:xx:xx:xx:xx:xx, proto UDP, 192.168.0.1:123->192.168.0.30:123, len 56 
13:17:57 firewall,info input: in:br1 out:(none), src-mac xx:xx:xx:xx:xx:xx, proto UDP, 192.168.0.1:123->192.168.0.30:123, len 56 
13:17:59 firewall,info input: in:br1 out:(none), src-mac xx:xx:xx:xx:xx:xx, proto UDP, 192.168.0.1:123->192.168.0.30:123, len 56 
13:18:01 firewall,info input: in:br1 out:(none), src-mac xx:xx:xx:xx:xx:xx, proto UDP, 192.168.0.1:123->192.168.0.30:123, len 56

In NTP Client window I observed transition from started state to reached state, but no any further progress in 5 minutes and more. Have tried to improve logging by adding debug and ntp topics in Logging window, but no more log information yielded.
Any ideas what’s I’ve overlooked?

Here is a list of public NTP servers (with ips!):
http://tf.nist.gov/tf-cgi/servers.cgi
I tried 208.66.175.36 and it works well. If this works, then it may be your Windows NTP server setup.

Thanks, it is possible that I have some misconfiguration, but currently my RB750 is limited to local access only, thus I have to tie to something that at least has a battery.

That is a great idea! That is exactly the setup I use, but I use a Linux NTP server for all my routers. However, that does not determine where the error is occurring. Can you allow the 750 through to that ip just for the NTP test? Otherwise, since it works well here with V4.14, I presume it is your WindowsXP NTP server setup.

No, I can’t check whether RB750 will gather info from external NTP server, because I am behind a proxy server.

I have a local linux workstation as well and tried it as a NTP server. Unfortunately in this case RB750 passed only one packet and gather one reply and moves to failed state. I suspect that this is because this linux workstation does not have a reference server (i.e. it does not get info from upstream NTP server). One more interesting thing: Rb750 makes requests using NTP v4. Windows workstation answers with NTP v3 (all 10 packets), linux replies with NTP v4.

Have tried to connect RB750 to WinXP NTP server without reference NTP Server. The result expected: failed. RB750 does not want to sync with NTP servers which does not have reference server.

Thus all observations now show that NTP client implemented in RB750 does not understand NTP answers from NTP servers implementing NTP v3.

Proxy server is only redirecting tcp port 80. This is udp port 123. Does the proxy server firewall intercept all ports?

The proxy is HTTP/HTTPS and SOCKS server. UDP 123 port is occupied on the server, so I can’t simply forward it.