|
1. NTP Introduction
2. NTP client / server introduction
3. Showing the time
4. Setting the time manually
5. Setting the time automatically
6. Maintening the NTP server
7. Installing NTP tools
8. Related documents
9. About
1.1. Scope
This howto briefly describes the main possibilities to keep your system time up-to-date using NTP.
1.2. NTP
The Network Time Protocol (NTP), has been created to transport time information, so that hosts can easily synchronize their clocks to thrusted time servers.
It has a high resolution: about 2^-32 seconds, which correpons to less than 1 ns.
Before proceeding, it is better to clarify some important points.
1.3. Hardware clock
This is the time provided by the on board RTC.
When your PC is turned off, there is a chip on the motherboard that still provides the time. It is powered by a little battery and still works if you unplug the power cabel.
1.4. Software clock
This is the time provided by the kernel, also known as system clock.
During the boot, it reads the correct time from the hardware clock and while the system has beeing shutting down it writes the up-to-date time to the hardware clock.
1.5. GMT
The Greenwich Meridian Time (GMT) is the time at the Royal Observatory in Greenwich (London).
It is based on the heart rotation around the sun, which introduces a little error every year.
In the past, it was the main reference for a world time, but due to this drift error, it has been replaced by UTC.
1.6. UTC
The Universal Time Coordinated (UTC) is an atomic time scale used to indicate a time valid in the whole world at the same moment. In fact, it is a world's standard time.
It is called "coordinated" because there are several institutions that estimate the current time and UTC is the combination of these.
It is not subject to Daylight Saving Time or other changes.
Setting the hardware clock to UTC is highly recommended. You can change this behaviour setting the variable
UTC
in the file
/etc/default/rcS
and rebooting the system.
Using UTC your system clock is calculated based on your timezone, which is stored in the file
/etc/timezone
and automatically changed to Daylight Saving Time and back to normal time on the correct moment.
You can change your timezone by calling
tzconfig
1.7. SNTP
The Simple Network Time Protocol (SNTP) is an easier implementation of the NTP.
It misses some internal algorithms that are not needed for some kind of server.
1.8. XNTP
XNTP was an experimental version of NTP, which has been dropped in NTPv4.
2.1. How to they work
Both NTP client and NTP server ask for the time to a remote NTP server and automatically update their system clock (inclusive the hardware clock!). There is just a difference about how they do that.
Optionally, a NTP daemon, once synchronized, is able to provide the time to other clients. These clients can be either NTP clients or NTP servers.
2.2. Sync possibilities
Basically, there are two ways to keep your system clock up-to-date using NTP:
- using a client
- using a server
The best solution is to install a server, also if you will NOT provide time information to other hosts!
To understand why, please read the following chapters.
2.3. NTP client
This solution does not use a daemon on your host. There are many tool to accomplish this task. Some of theme are:
- ntpdate
- rdate
- ...
The principle is simple: somebody has to call one of these clients in order to synchronize the clock. This can be done either manually (by command line) or with a cron job, which regularly calls them.
Also if you have a laptop, I suggest you to install a NTP server, which has more advantages (see next chapter).
2.4. NTP server
This solution uses a daemon on your host.
The daemon has a lot of advantages (see next chapters).
2.4.1. No any sudden change
If the daemon has to correct the system clock, it does it slowly. This because there could be some service or application which looks for the time to accomplish some task and if the system clock changes suddenly they could have unexpected behaviour.
One problem was for example with the cronjob. If the system clock was suddenly changed from 2000-01-01 to 2007-05-28, it had to call 7 years of cron jobs, which could be fatal for the host!
This problem has been fixed in the cronjob, but could be present in other services or applications.
Therefore, using a daemon the risk of unexpected behaviour of the system is reduced.
2.4.2. Permanently in sync
The daemon, at the contrary of a cron job, can spontaneously ask for the current time.
Therefore, using a daemon the time will be permanently in sync.
2.4.3. Smart correction
Each hardware clock has a different drift error. The daemon tries to calculate this drift and in case of no connection with remote time servers it can estimate the current time considering this drift error.
Therefore, using a daemon the time will be more accurate in case of connection lost.
3.1. Command "date"
3.1.1. Installing
This command is provided by the Debian package coreutils. Therefore, to install it run
apt-get update
apt-get install coreutils
3.1.2. Showing local time
This command can be used to display the current system time.
To use it, simply call
date
It returns something like
Mon May 28 11:01:01 CEST 2007
CEST means, my system time is set to Central Europe.
3.1.3. Showing universal time
This command can be used to display the current UTC time (which should normally be used by your hardware clock):
date -u
It returns something like
Mon May 28 09:01:01 UTC 2007
How you can see, at the 28th of May, my system time (local time) is set to my UTC time + 2 hours.
3.2. Command "rdate"
3.2.1. Installing
This command is provided by the Debian package rdate. Therefore, to install it run
apt-get update
apt-get install rdate
3.2.2. Showing remote time
This command can be used to display the current system time of a remote host.
To use it, call somethink like:
rdate -p tempo.ien.it
It returns something like
Mon 28 May 2007 11:13:38 AM CEST
4.1. Command "date"
4.1.1. Installing
This command is provided by the Debian package coreutils. Therefore, to install it run
apt-get update
apt-get install coreutils
4.1.2. Setting system time
This command can be used to set the system time.
To use it, simply call something like
date --set 2007-05-28
date --set 19:02:30
4.2. Command "rdate"
4.2.1. Installing
This command is provided by the Debian package coreutils. Therefore, to install it run
apt-get update
apt-get install coreutils
This command can be used to synchronize the system time with a remote NTP server.
To use it, call something like:
rdate tempo.ien.it
Note that rdate has an accuracy of only 1 second.
4.3. Command "ntpdate"
This command can be used to synchronize the system time with a remote NTP server.
To use it, simply call
ntpdate ipaddr
where ipaddr is the address of a NTP server.
To display some more information, call
ntpdate -d ipaddr
where ipaddr is the address of a NTP server.
4.4. Command "clock"
This command is obsolete. Use hwclock instead (see next chapter).
This command can be used to synchronize the hardware time with the system time.
To use it, simply call
clock -w
If hardware clock is in UTC (like it should be!), call:
clock -w -u
4.5. Command "hwclock"
This command can be used to synchronize the hardware time with the system time.
To use it, simply call
hwclock --systohc
If hardware clock is in UTC (like it should be!), call:
hwclock --systohc --utc
4.6. Command "time"
This is a stopwatch and has nothing to do with the clock!
I have reported it, just because it is simple to think that this command could have something to do with the system/hardware clock.
To see what it can do, try this command:
time sleep 5
You will get something like
real 0m5.008s
user 0m0.000s
sys 0m0.000s
that indicates how many seconds have been necessary to execute the specified command (in this case the "sleep" command).
5.1. Introduction
This chapter explains how to change the time automatically using a NTP server.
How already explained, there are other possibilities to do this, but since they are not the best choice, I don't treat them in this document.
5.2. Installing a NTP server
In Debian, the NTP server can be installed in this way:
apt-get update
apt-get install ntp
Once installed, it is automatically started.
5.3. Configuring a NTP server
In Debian, the default configuration file will already synchonize your NTP server with several remote NTP servers and provide time to every body asks it for.
The server has just one configuration file called
/etc/ntp.conf
I have reported here the default config file:
# /etc/ntp.conf, configuration for ntpd
driftfile /var/lib/ntp/ntp.drift
statsdir /var/log/ntpstats/
statistics loopstats peerstats clockstats
filegen loopstats file loopstats type day enable
filegen peerstats file peerstats type day enable
filegen clockstats file clockstats type day enable
# You do need to talk to an NTP server or two (or three).
#server ntp.your-provider.example
# pool.ntp.org maps to more than 300 low-stratum NTP servers.
# Your server will pick a different set every time it starts up.
# *** Please consider joining the pool! ***
# *** <http://www.pool.ntp.org/join.html>; ***
server 0.debian.pool.ntp.org iburst
server 1.debian.pool.ntp.org iburst
server 2.debian.pool.ntp.org iburst
server 3.debian.pool.ntp.org iburst
# By default, exchange time with everybody, but don't allow configuration.
# See /usr/share/doc/ntp-doc/html/accopt.html for details.
restrict -4 default kod notrap nomodify nopeer noquery
restrict -6 default kod notrap nomodify nopeer noquery
# Local users may interrogate the ntp server more closely.
restrict 127.0.0.1
restrict ::1
# Clients from this (example!) subnet have unlimited access,
# but only if cryptographically authenticated
#restrict 192.168.123.0 mask 255.255.255.0 notrust
# If you want to provide time to your local subnet, change the next line.
# (Again, the address is an example only.)
#broadcast 192.168.123.255
# If you want to listen to time broadcasts on your local subnet,
# de-comment the next lines. Please do this only if you trust everybody
# on the network!
#disable auth
#broadcastclient
If you change something in the configuration file, remember to restart the daemon:
/etc/init.d/ntp restart
6.1. Introduction
There are a lot of tools that can be used to analyze the status of a NTP daemon.
6.2. Checking log file
The first place to look at in case of problem is the configuration file.
By default, NTP logs to syslog. Therefore, to filter NTP informations:
grep ntpd /var/log/syslog
6.2.1. Logging to /var/log/ntpd.conf
In case of NTP problem, I suggest you to log to a separate file. This case be done by adding the following line in your configuration file:
logfile /var/log/ntpd.log
and restarting the daemon
/etc/init.d/ntp restart
Now, to look for NTP logs, call
tail -f -n1000 /var/log/ntpd.log
6.2.2. Messages for stopping, starting and restarting the daemon
If the daemon has been stopped, you should see something like:
May 28 12:09:18 toby ntpd[13623]: ntpd exiting on signal 15
If the daemon has been started, you should see something like:
May 28 12:09:43 toby ntpd[14049]: ntpd 4.2.2p4@1.1585-o Sun Mar 4 13:21:35 UTC 2007 (1)
May 28 12:09:43 toby ntpd[14050]: precision = 1.000 usec
May 28 12:09:43 toby ntpd[14050]: Listening on interface wildcard, 0.0.0.0#123 Disabled
May 28 12:09:43 toby ntpd[14050]: Listening on interface wildcard, ::#123 Disabled
May 28 12:09:43 toby ntpd[14050]: Listening on interface lo, ::1#123 Enabled
May 28 12:09:43 toby ntpd[14050]: Listening on interface eth1, fe80::20e:a6ff:fe31:aacd#123 Enabled
May 28 12:09:43 toby ntpd[14050]: Listening on interface lo, 127.0.0.1#123 Enabled
May 28 12:09:43 toby ntpd[14050]: Listening on interface eth1, 192.168.128.110#123 Enabled
May 28 12:09:43 toby ntpd[14050]: kernel time sync status 0040
May 28 12:09:43 toby ntpd[14050]: frequency initialized 58.971 PPM from /var/lib/ntp/ntp.drift
May 28 12:09:53 toby ntpd[14050]: synchronized to 216.168.105.34, stratum 3
May 28 12:09:53 toby ntpd[14050]: kernel time sync enabled 0001
Note that the server has not been synchronized until you see the following line:
kernel time sync enabled 0001
and this can take up to 15 minutes!
The NTP server will not provide time to NTP clients until its system time has been synchronized with at least 1 remote NTP server!
@EMI: add here error message of this case
6.3. Command "ntpdc"
This is a NTP query program.
It can be used on your NTP server to check if it has been synchronized with one ore more remote NTP servers.
Type the following command:
ntpdc -p
You will get something like
remote local st poll reach delay offset disp
=======================================================================
*chiccp.net 192.168.128.110 2 64 377 0.12799 0.000470 0.05133
=168.75.65.20 192.168.128.110 3 64 377 0.11221 0.000187 0.04698
=osource.org 192.168.128.110 3 64 377 0.14583 -0.002432 0.06522
=192.135.48.21 192.168.128.110 2 64 377 0.14502 0.002252 0.06416
The sign before the remote NTP server name or IP address (first column) indicates if the NTP server is in sync or not.
The sign "=" indicates not in sync.
The sign "*" indicates in sync.
I guess, there is always only one remote NTP server to which your NTP server is synchronized.
You can use this command to get some additional information. Check these:
ntpdc -c loopinfo
and
ntpdc -c kerninfo
6.4. Command "ntpdate"
This command, already explained in the previous chapters, can be used to check if your NTP server provides the time.
Unfortunately, you cannot run it on the same host running your NTP server, because both NTP server and client use the same socket and the socket is already locked by the NTP server.
Therefore, to make this test, you must have a second GNU/Linux host where you can run this command.
For information about how to use it, refer you to the previous chapters.
6.5. Command "ntptrace"
This command follows the chain starting from the given server back to the the stratum 1 server. In this ways, it is possible to know from what base server your time is indirectly synchronized.
To use it, symple call
ntptrace NtpServerUsedByYourNtpServer
For example, supposing you get your NTP information from pool.ntp.org, calling
ntptrace pool.ntp.org
will give you something like
pool.ntp.org: stratum 2, offset 0.036120, synch distance 0.078492
time-C.timefreq.bldrdoc.gov: stratum 1, offset 0.000000, synch distance 0.002650, refid 'ACTS'
@EMI: check if really so!
This chapter explains how to install in Debian GNU/Linux all the tools described in this document.
7.1. Installing command "clock"
This command is obsolete and is not anymore provided by Debian. Use hwclock instead.
7.2. Installing command "date"
This command is provided by the Debian package coreutils. Therefore, to install it run
apt-get update
apt-get install coreutils
7.3. Installing command "hwclock"
This command is provided by the Debian package util-linux. Therefore, to install it run
apt-get update
apt-get install util-linux
7.4. Installing the daemon "ntp"
This daemon (correspondig to the NTP server/client) is provided by the Debian package ntp. Therefore, to install it run
apt-get update
apt-get install ntp
7.5. Installing command "ntpdate"
This command is provided by the Debian package ntpdate. Therefore, to install it run
apt-get update
apt-get install ntpdate
7.6. Installing "rdate"
This command is provided by the Debian package rdate. Therefore, to install it run
apt-get update
apt-get install rdate
http://www.ntp.org
http://tldp.org/LDP/sag/html/index.html
http://www.debian.org/doc/manuals/system-administrator/ch-sysadmin-time.html
9.1. Author
Emidio Planamente <eplanamente@gmx.ch>
9.2. Feedback
Any feedback is welcome.
9.3. History
Version 1.0
First public release
|
