Network Time Protocol

From ArchWiki
Revision as of 17:17, 11 February 2011 by Kynikos (Talk | contribs) (Updated section 1.2 and improved formatting)

Jump to: navigation, search

This template has only maintenance purposes. For linking to local translations please use interlanguage links, see Help:i18n#Interlanguage links.


Local languages: Català – Dansk – English – Español – Esperanto – Hrvatski – Indonesia – Italiano – Lietuviškai – Magyar – Nederlands – Norsk Bokmål – Polski – Português – Slovenský – Česky – Ελληνικά – Български – Русский – Српски – Українська – עברית – العربية – ไทย – 日本語 – 正體中文 – 简体中文 – 한국어


External languages (all articles in these languages should be moved to the external wiki): Deutsch – Français – Română – Suomi – Svenska – Tiếng Việt – Türkçe – فارسی

This article describes two different methods to synchronize the Linux software clock (system clock) to internet time servers. For more information on computer timekeeping, see Time. Traditional ntpd is explained first and then OpenNTPD (part of the OpenBSD project) is explained. Both daemons can sync the local clock and act as a time server if needed. OpenNTPD is designed to be simple and secure, while ntpd has a larger set of features. Note that OpenNTPD is not currently maintained for Linux (see this thread).

ntpd

Installation

ntp is available from [extra]:

Template:Cli

Configuration

The first thing you define in your Template:Filename is the servers your machine will synchronize to. NTP servers are classified in a hierarchical system with many levels called strata: the devices which are considered independent time sources are classified as stratum 0 sources; the servers directly connected to stratum 0 devices are classified as stratum 1 sources; servers connected to stratum 1 sources are then classified as stratum 2 sources and so on.

It has to be understood that a server's stratum cannot be taken as an indication of its accuracy or reliability. Tipically, stratum 2 servers are used for general synchronization purposes: if you don't already know the servers you're going to connect to, you should use the pool.ntp.org servers (alternate link) and choose the server pool that is closest to your location.

The following lines are just an example:

server 0.it.pool.ntp.org iburst
server 1.it.pool.ntp.org iburst
server 2.it.pool.ntp.org iburst
server 3.it.pool.ntp.org iburst

The iburst option is recommended, and sends a burst of packets if it cannot obtain a connection with the first attempt. The burst option should never be used without explicit permission and will likely result in blacklisting.

If you're setting up a ntp server, you need to add localhost as a server, so that, in case it loses internet access, it won't stop serving time to the network; add localhost as a stratum 10 server (using the fudge command) so that it will never be used unless internet access is lost:

server 127.127.1.0
fudge  127.127.1.0 stratum 10

The next thing you have to do is add the drift file (which keeps track of your clock's time deviation) and optionally the log file location:

driftfile /var/lib/ntp/ntp.drift
logfile /var/log/ntp.log

Now all that's left to do is define the rules that will allow clients to connect to your service (localhost is considered a client too) using the restrict command; you should already have a line like this in your file:

restrict default nomodify nopeer

This restricts everyone from modifying anything and prevents everyone from querying your time server.

You can also add other options:

restrict default kod nomodify notrap nopeer noquery

In the past, notrust option was used too, but its function has changed to mean that authentication with a key is required.

Following this line, you need to tell ntpd what to allow through into your server; the following line is enough if you're not configuring a ntp server:

restrict 127.0.0.1

Otherwise you can add more clients like in this example:

restrict 1.2.3.4 nomodify
restrict 192.168.0.0 mask 255.255.255.0 nomodify notrap

This tells ntpd that 1.2.3.4 and all IP addresses from the 192.168.0.0 range will be allowed to synchronize on this server, but they will not be allowed to modify anything. All other IP addresses in the world will still obey the default restrictions (the first line in the Template:Filename).

In the end, for the most basic configuration, the complete file will look like this (all comments have been stripped out for clarity):

Template:File

Actually, defining the log file is not mandatory, but it's always a good idea to have a feedback for ntpd operations.

Lastly, never forget man pages: Template:Codeline is likely to answer any doubts you could still have.

Template:Gentoo

Running as a daemon

Starting ntpd

Add ntpd to the DAEMONS array in Template:Filename to have it start at boot:

DAEMONS=(syslog-ng network @ntpd ...)
NetworkManager

ntpd can be brought up/down along with a network connection through the use of NetworkManager's dispatcher scripts. You can install the needed script from [community]:

Template:Cli

Running as non-root user

When compiled with --enable-linux-caps, ntp can be run as a non-root user for increased security (the vanilla Arch Linux package has this enabled).

Note: Before attempting this, make sure ntp has already created Template:Filename.

Create ntp group and ntp user:

Template:Cli

Change ownership of ntp directory to the ntp user/group:

Template:Cli

Edit Template:Filename and change

NTPD_ARGS="-g"

to

NTPD_ARGS="-g -u ntp:ntp"

Syncing the clock without running the daemon

If what you want is just synchronize your system clock at boot time without running ntpd as a daemon, you can add this line to your Template:Filename:

ntpd -qg &
Note:
  • In order for this method to work you have to make sure that, when Template:Filename is executed, the network connection has already been initialized (for example you shouldn't background essential network-related daemons in Template:Filename)
  • The hardware clock is already automatically realigned with the system clock each time you halt the computer, through a command in Template:Filename
Warning:
  • Using this method is highly discouraged on servers and in general on machines that need to run continuously for more than 2 or 3 days, as the system clock will be updated only once at boot time.
  • Running "ntpd -qg" as a cron event is to be completely avoided, unless you are perfectly aware of how your running applications would react to instantaneous system time changes.

OpenNTPD

Installation

OpenNTPD can be installed from [community]:

# pacman -S openntpd

The default configuration is actually usable if all you want is to sync the time of the local computer. For more detailed settings, the Template:Filename file must be edited:

To sync to a particular server, uncomment and edit the "server" directive. You can find the server's URL in your area at www.pool.ntp.org/zone/@.

server ntp.example.org

The "servers" directive works the same as the "server" directive, however, if the DNS name resolves to multiple IP address, ALL of them will be synced to. The default, "pool.ntp.org" is working and should be acceptable in most cases.

pool.ntp.org

Any number of "server" or "servers" directives may be used.

If you want the computer you run OpenNTPD on to also be a time server, simply uncomment and edit the "listen" directive.

For example:

listen on *

will listen on all interfaces, and

listen on 127.0.0.1

will only listen on the loopback interface.

Your time server will only begin to serve time after it has synchronized itself to a high resolution. This may take hours, or days, depending on the accuracy of your system.

If you would like to run OpenNTPD at boot, add Template:Codeline the DAEMONS variable in your Template:Filename following your network daemon.

DAEMONS=(syslog-ng network openntpd ...)

If openntpd is being used to set local system time only, it may be safely backgrounded.

DAEMONS=(syslog-ng network @openntpd ...)

To see the status of NTP syncing, visit Template:Filename and look for entries with "ntpd".

OpenNTPD adjusts the clock by small amounts at a time. It is designed this way to prevent sudden, large time fluctuations in your system, which could adversely affect system services (e.g., cron jobs). Thus, it can take some time to correct the time.

If your clock is off by more than 180 seconds you can try "Template:Codeline" in the console. If ntpd is already running, you can simply restart it with Template:Codeline, as the Arch openntpd package uses the "-s" flag by default. See Template:Codeline for more info. You can also set the system clock to as close to possible to the actual time and then let OpenNTPD fine tune the time.

Making openntpd dependent upon network access

If you have intermittent network access (you roam around on a laptop, you use dial-up, etc), it does not make sense to have Template:Codeline running as a system daemon on start up. Here are a few ways you can control Template:Codeline based on the presence of a network connection. These instructions should also work for Template:Codeline found further below.

Using netcfg

If you are using netcfg, you can also start/stop openntpd as a POST_UP/PRE_DOWN command in your network profile:

POST_UP="/etc/rc.d/openntpd start || true"
PRE_DOWN="/etc/rc.d/openntpd stop || true"

Of course, you will have to specify this manually for each network profile.

Using NetworkManager dispatcher

OpenNTPD can be brought up/down along with a network connection through the use of NetworkManager's dispatcher scripts. You can install the needed script from [community]:

# pacman -S networkmanager-dispatcher-openntpd

Using wicd

These instructions require wicd 1.7.0 or later, which is available in the standard Arch repository. You will also need write access to Template:Filename.

Note: Remember to make these two scripts executable using Template:Codeline

Make one shell script inside Template:Filename with the following:

#!/bin/sh
/etc/rc.d/openntpd start

Similarly, make another shell script inside Template:Filename with the following:

#!/bin/sh
/etc/rc.d/openntpd stop

Using dhcpcd hooks

Another possibility is to use dhcpcd hooks to start and stop openntpd. When dhcpcd detects a change in state it will run the following scripts:
Template:Filename
Template:Filename
Template:Filename

The following example uses Template:Filename to start and stop openntpd depending on dhcp status:

[ "$interface" != "eth0" ] && exit 0

if $if_up; then
    pgrep ntpd &> /dev/null || /etc/rc.d/openntpd start
elif $if_down; then
    pgrep ntpd &> /dev/null && /etc/rc.d/openntpd stop
fi

Troubleshooting

Error adjusting time

If you find your time set incorrectly and in log you see:

openntpd adjtime failed: Invalid argument

Try:

ntpd -s -d

This is also how you would manually sync your system.

Increasing time shift

Starting openntpd in the background could lead to synchronization errors between the actual time and the time stored on your computer. If you recognize an increasing time difference between your desktop clock and the actual time, try to start the openntpd daemon normal and not in the background.

Initialization Failure

Openntpd may fail to initialize properly if it is started before the network is fully configured. In some cases you may want to remove Template:Codeline from the DAEMONS array in Template:Filename and add the following line to Template:Filename:

(sleep 300 && /etc/rc.d/openntpd start) &
Note: This method is an alternative to the four methods listed above. The other three methods are preferred and work better. Use this as a last resort.

This will wait 5 minutes before starting openntpd, which should give the system sufficient time to set up the network properly. If your network settings change often, you may also consider restarting the daemon regularly with cron.

External links