Re: Is port 37 safe to let out?

From: Menno Duursma (pan_at_desktop.lan)
Date: 06/07/05


Date: Tue, 07 Jun 2005 12:07:15 +0200

On Mon, 06 Jun 2005 20:54:30 -0700, Mikhail Zotov wrote:
> Menno Duursma wrote:
>> On Sun, 05 Jun 2005 21:41:41 -0700, Anthony Ewell wrote:
>> > matt_left_coast wrote:
>> >> Anthony Ewell wrote:
>>
>> >>> I am getting a lot of port 37 (time) outbound connection attempts
>> >>> on my iptables firewall.

I'd be interesting to find out: how come?

>> >>> Is it safe to "let it (port 37) out?"
>>
>> Maybe,

Probably.

>> but i wouldn't.

Unless for some reason you need this to work (in which case, you may want
to restrict it some.)

Apperently some box behind it your firewall wants to know how the clock is
set on some outsite host for some reason. And trys to use the RFC868 Time
Protocol to do so. Better to just sync one or two server boxen to
pool.ntp.org or something, and have them provide "time" services for LAN
connected machines.

Who knows: maybe it's actually an attempt of one of your users/machines to
create/get a tunnel through your firewall.

>> Unless you are on a LAN and have some box setup with ntpd (or a cron
>> job running "ntpdate") which provides "time" broadcasts

s/broadcasts/services/

Sorry this may well be incorrect (although Google tells there are Time
implementations which can send/recive broadcasts, i don't see it in the
RFC.) But i might have been thinking BSD TSP (time synchronization
protocol) here. Which uses UDP port 525 instead:
http://www.linuxvalley.it/encyclopedia/ldp/manpage/man8/timed.8.php

My bad.

Thanks for pointing out my error Mikhail.

>> to the subnet you're on, for other machines - yours - to sync with.

Which you (cron) would do with "netdate" probably.

>> >> Outbound means you are trying to connect to a time server on the
>> >> internet.
>>
>> No it doesn't.

Well, your correct (ofcource) sorry again. However a "time server on the
internet" would generally be providing an (S)NTP service, rather then the
Time Protocol one. As the former is much more accurate.

>> >> If you want to sync to the atomic clocks, you need to let the packet
>> >> out.
>>
>> No you don't (or atleast not unless you have some atomic clocks within
>> your subnet/broadcast domain providing a time service to you
>> (unlikely.))
>
> Menno, could you please explain your point.

The basic "time" service isn't very accurate, unless both client and
server know about some extension to it. And even then, only when used over
a relatively lo-latency (LAN) network.

-- 
-Menno.


Relevant Pages

  • Re: ssh protocol 2,1
    ... That means both client and server are both ssh 1. ... I thought Protocol 1 had a security flaw. ... There must be a reason for allowing this vulnerability by default. ...
    (RedHat)
  • Re: Should DCs with DNS point to self first?
    ... > when you have all locally, by doing so IMO you're wasting server ... > good reason to do so IMO. ... there are far more issues associated with pointing a DC at itself for primary DNS than pointing at something else. ...
    (microsoft.public.windows.server.active_directory)
  • Re: Recommended IMAP server
    ... Is there be any particular reason I should swap? ... pop3 mailboxes and passing it on to the imap server. ... > Trying to strip the HTML mail is quite a bit of a mess. ... tend to be much more attached to email clients. ...
    (comp.os.linux.networking)
  • Re: Lessons! :)
    ... there's an herbal protocol that may be helpful. ... And there's a website where people can ask questions, and Buhner ... and may be one reason why some of us seem unable ... Those of us with chronic Lyme know we're a ...
    (rec.equestrian)
  • Re: Backup Failing
    ... Directory Service Access auditing is enabled. ... and then click Server Management. ... Click Edit to open Group Policy Object Editor. ... >>> Reason: The process cannot access the file because it is being used by ...
    (microsoft.public.windows.server.sbs)