[2024-feb-29] Sad news: Eric Layton aka Nocturnal Slacker aka vtel57 passed away on Feb 26th, shortly after hospitalization. He was one of our Wiki's most prominent admins. He will be missed.

Welcome to the Slackware Documentation Project


This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision
Previous revision
howtos:network_services:ntp [2014/02/17 21:13 (UTC)] kikinovakhowtos:network_services:ntp [2022/09/20 19:46 (UTC)] (current) – [Sources] tag to Slackware 15.0 and additional author metaed
Line 2: Line 2:
 ====== Synchronize your network with NTP ====== ====== Synchronize your network with NTP ======
-NTP (//Network Time Protocol//) allows clock synchronization between computer systemsThe following HOWTO describes:+NTP is the //Network Time Protocol//, used to synchronize host clocks to one 
 +another. Your Slackware distribution comes with NTP preinstalled.
-  * configuring an NTP server on Slackware Linux; +Your reasons for running NTP might include:
-  * synchronizing client PCs with your local NTP server.+
-===== Introduction =====+  * make timestamps in system logs agree with one other, to make sense of events recorded in multiple system logs 
 +  * enable software protocols and encryption that depend on accurate time, e.g., Kerberos or PCI 
 +  * prevent software build issues caused when your remote filesystem says your source file was modified in the future 
 +  * prevent issues with database software that cannot tolerate setting the host clock back in time 
 +  * know when to stop hacking and turn on the new //Star Trek: Picard//
-When several users manipulate shared data on different client PCs on a networkit's important that these machines are all synchronized. This is especially true if you share files over NFS, or if you use NIS for centralized authentication. You'll get all sorts of weird errors if your clocks are out of sync. Unfortunately, the clients' onboard clocks aren't sufficiently precise.+Using NTP, your Slackware host can become any of the following:
-That's where NTP (//Network Time Protocol//) comes in handy. It allows networked machines to adjust their clocks so as to be perfectly synchronized. series of public time servers on the Internet allow the reception of the exact time. From this pointwe can use NTP in several ways.+  * An NTP //client//, that corrects its own host clock to match that of another host 
 +  * //standalone// NTP client, that does not match its own clock to another host but does take advantage of NTP's ability to make frequency corrections when the host clock gains or loses time too quickly 
 +  * An NTP //server//that shares its system time with other hosts 
 +  * A //primary// NTP server, that gets its system time not from another host but from a hardware clock that has direct access to the Coordinated Universal Time timekeeping process (also known as a //stratum 1 host//)
-  * The ''ntpdate'' command makes an initial correction of the BIOS clock. +It is very common to operate NTP as both a server and a client. In other 
-  * This one-time-adjustment isn't sufficient for a server that is supposed to be up 24/7, since its clock will drift away gradually from the exact time. In that casewe have to configure the ''ntpd'' daemon (shipping with the ''ntp'' package). This daemon contacts public time servers at regular intervals and proceeds with incremental corrections of the local clock. +wordsyour Slackware host can get accurate time from another host on the 
-  * The ''ntpd'' daemon can in its turn be configured as a time server for the local client machines.+Internet and then serve that time to hosts on your local network.
-It's considered good practice to use ''ntpdate'' for the initial adjustment and ''ntpd'' for regular time synchronization.+<note important> 
 +This HOWTO is based on Slackware version 15.0 and the NTP reference 
 +implementation version 4.2.8 that comes with it. To check your Slackware 
 +version see: "[[howtos:misc:checking_a_slackware_version|Checking a Slackware 
 +Version]]". To check your NTP software version, use the ntpd command:
-===== Firewall considerations =====+  $ ntpd --version 
 +  ntpd 4.2.8p15@1.3728-o Fri May 21 19:02:16 UTC 2021 (1) 
-The NTP services uses UDP port 123. Open this port if you want to allow remote machines to connect to your NTP server.+===== Access control =====
-===== Synchronize a LAN server or a public root server with an NTP server on the Internet =====+NTP uses UDP port 123. Open port 123 in your host firewall if you want to 
 +allow other hosts to connect to your host. Open port 123 in your Internet 
 +firewall if you want access to Internet time.
-Create an empty log file:+NTP uses the ''restrict'' command in ''/etc/ntp.conf'' to impose additional 
 +restrictions by creating an ACL (//access control list//). The ACL is used by 
 +a mini-firewall within NTP itself that drops inbound packets based on options 
 +you choose.
-<code> +Please turn your attention to the ACL pre-supplied by Slackware in 
-# touch /var/log/ntp.log +''/etc/ntp.conf'':
-Visit [[http://www.pool.ntp.org|http://www.pool.ntp.org]] and choose a list of servers according to your country.+  restrict default limited kod nomodify notrap nopeer noquery 
 +  restrict -6 default limited kod nomodify notrap nopeer noquery 
 +  restrict 127.0.0.
 +  restrict ::1
-Configure the NTP service by editing ''/etc/ntp.conf''You might backup the existing ''ntp.conf'' file and start from scratch+This ACL prevents NTP from running as either a client (nopeer) or a server 
 +(noquery)It drops all packets except requests for basic information (first 
 +two lines)It makes an exception for packets that originate from your 
 +Slackware host itself (last two lines). This exception is what lets you 
 +control your own NTP service using the ''ntpq'' command.
-In the example below, the list of four servers is chosen for my company's location (France):+The second line is redundant and should be deleted.
-<code> +If you want to use the public NTP server pool, you must add a line to relax 
-# /etc/ntp.conf+the ACL restrictions enough to allow peering with associationsAdd a 
 +''restrict source'' command without the ''nopeer'' flag, such as:
-driftfile /etc/ntp/drift +  restrict source limited kod nomodify notrap noquery
-logfile /var/log/ntp.log+
-server 0.fr.pool.ntp.org +If you want to allow devices on your network (or anywhere you like) to get 
-server 1.fr.pool.ntp.org +time from this host, you must add a line to relax these restrictions to permit 
-server 2.fr.pool.ntp.org +clientsAdd a ''restrict //address//'' command that identifies the device(s) 
-server 3.fr.pool.ntp.org+and/or network(s) that are allowed to get timeIf your local network is 
 +'''', you could add:
-server +  restrict mask limited kod nomodify notrap nopeer
-fudge stratum 10+
-restrict default nomodify nopeer notrap +To let you control your NTP service from your maintenance VLAN and not just 
-restrict mask +the host itself, you might want to add the VLAN with no restrictions. 
-</code>+Supposing the maintenance VLAN is '''', you could add:
-Here's a little explanation for some options:+  restrict mask
-  * The ''fudge stratum 10'' directive is a "dummy" server acting as fallback IP in case the external time source becomes momentarily unreachableWhen this happens, NTP will continue to work and base itself on this "internal" server+If you want more sophisticated access control than what's described here, for 
-  * NTP has its own arsenal of rules to limit access to the service, which can be used independently from a firewall. The ''restrict'' directives in the above configuration prevent distant computers from changing the servers' configuration (first ''restrict'' statement), and the machine is configured to trust itself (second ''restrict'' statement).  +example to encrypt traffic or let you authenticate for administration tasks 
-  * A ''restrict'' statement without any argument but followed by the hostname boils down to an ''allow all''.+from any host, look into the secure authentication features of NTPSee: 
 +Support]]" at //[[http://doc.ntp.org|The NTP Project]]//.
-==== Manage the NTP service ====+===== Diagnostic logging =====
-Before starting the service, proceed to an initial adjustment of your system clock:+NTP prefers to use SYSLOG for logging. There is an alternate logging feature 
 +in NTP itself that can be used instead. Looking again at the preinstalled 
 +Slackware ''/etc/ntp.conf'', the alternate logging feature has already been 
 +turned on:
-<code+<note important
-# ntpdate pool.ntp.org +  logfile /var/log/ntp 
-</code>+Not recommended. If you use the alternate logging feature, you are also 
 +responsible for managing the logfile so that it does not eventually consume 
 +all the available space in the filesystem. Not having to do this is one of the 
 +big advantages of using SYSLOG
-<note>The ''ntpdate'' command is normally considered obsolete, but it still comes in handy when performing important time adjustmentsThe "orthodox" way would be to use the ''ntpd -g'' command - the official replacement for ''ntpdate'' - but its use will fail if your system clock is off for more than half an hour.</note>+It is simpler to delete the ''logfile'' line and use Slackware's preinstalled 
 +SYSLOG packageUsing SYSLOG, NTP logs warnings and errors to 
 +''/var/log/syslog'', and routine status messages to ''/var/log/messages''.
-Activate the NTP service:+If you still want to use the alternate logging feature, be sure to create the 
 +empty file and make it writable by the NTP daemon:
-<code> +  touch /var/log/ntp.log 
-chmod +x /etc/rc.d/rc.ntpd +  # chown ntp:ntp /var/log/ntp.log
-Manage the NTP service:+==== Filtering the log ====
-<code> +NTP lets you filter certain messages out of the log, based on the message's 
-# /etc/rc.d/rc.ntpd start|stop|restart|status +''class'' and ''type''Currently there are four classes defined:
-Now display the list of servers your machine is actually connected to:+  clock peer sync sys
-<code> +and four types defined:
-# ntpq -p +
-remote           refid      st t when poll reach   delay   offset jitter +
-============================================================================== +
-*panopea.unstabl  2 u   30   64  377   56.136  -249.48 80.680 +
-+88-190-17-126.r   2 u   29   64  377   77.571  -205.94 94.278 +
-+      2 u   29   64  377   77.097  -249.57 85.641 +
--ntp.univ-poitie   3 u   29   64  377   57.747  -191.58 107.002 +
-LOCAL(0)        .LOCL.          10 l  164   64  374    0.000    0.000 0.001 +
-The little ''*'' asterisk preceding one of the above lines means your machine is effectively synchronized with the respective NTP server.+  info events status statistics
-<note tip>The first synchronization can take a few minutessometimes up to half an hour.</note>+Because the preinstalled Slackware ''/etc/ntp.conf'' does not customize the 
 +filteryou get out-of-box behavior. NTP will pass messages that are tagged 
 +with the ''sync'' class and drop all messages that are tagged with any other 
-===== Synchronize your client PC(swith your local NTP server =====+<note important> 
 +The out-of-box behavior unfortunately filters the message associated with at 
 +least one common fatal condition (termination of the ntpd process when the 
 +clock offset exceeds its panic threshold)
-In a LANit is considered good practice to synchronize only one machine - the server - with a public NTP server, and the client PCs with the local server. This saves bandwidth and takes some load off the public NTP servers.+If you want all available diagnostic messages loggedyou should disable all 
 +filtering by class or type in ''/etc/ntp.conf'':
-As above, proceed to an initial adjustment of the system clock:+  logconfig =allall
-<code> +===== Statistics gathering =====
-# ntpdate pool.ntp.org +
-Create an empty logfile:+NTP can keep a statistical record of its performance, that you can analyze to 
 +check the health of your NTP-managed clock. The preinstalled Slackware 
 +''/etc/ntp.conf'' already configures the directory path for these statistics:
-<code> +  statsdir /var/lib/ntp/stats
-# touch /var/log/ntp.log +
-Now configure NTP to synchronize with the LAN server. Replace the example's IP ('''') with your real server's IP:+But to actually collect statistics, you must create the empty directory and 
 +make it writable by the NTP daemon:
-<code> +  # mkdir /var/lib/ntp/stats 
-# /etc/ntp.conf+  chown ntp:ntp /var/lib/ntp/stats
-driftfile /etc/ntp/drift +and add a command to ''/etc/ntp.conf'' that identifies the statistics you want 
-logfile /var/log/ntp.log+collectedThe most commonly analyzed record is NTP's system clock updates in 
 +the ''loopstats'' file:
-server  statistics loopstats
-server +There are a total of eight recordtypes that NTP will keepFor information, 
-fudge stratum 10+see: 
 +Options]]" at //[[http://doc.ntp.org|The NTP Project]]//.
-restrict default ignore +At the end of this HOWTO, there is an example of charting the loopstats using 
-restrict mask +the preinstalled Slackware **gnuplot** package.
-restrict mask +
-  * The three ''restrict'' statements mean we're blocking all NTP traffic except for the client itself and the server.+<note important> 
 +As with the logfile, if you collect statistics, you are responsible for 
 +managing the statistics files so that they do not eventually consume all the 
 +available space in the filesystem. 
-Activate and start the NTP service:+===== Operating NTP as a client =====
-<code> +The preinstalled Slackware ''/etc/ntp.conf'' already has commands in it that 
-# chmod +x /etc/rc.d/rc.ntpd +would make NTP a client of the public NTP server pool, just commented out
-# /etc/rc.d/rc.ntpd start +Here are the relevant lines:
-As above, use the ''ntpq -p'' command to check if the synchronization went well:+<note important> 
 +  #server 0.pool.ntp.org iburst 
 +  #server 1.pool.ntp.org iburst 
 +  #server 2.pool.ntp.org iburst 
 +  #server 3.pool.ntp.org iburst 
 +As of NTP 4, this is no longer the recommended way to use the public NTP pool. 
 +You should replace the multiple server commands with a single ''pool'' 
 +command. The command that is equivalent to the lines above is: 
 +  pool pool.ntp.org 
 +Remember that you must also add the ''restrict source'' command to the ACL as 
 +described in an earlier section for this to work. 
 +<note important> 
 +The NTP Pool Project formerly recommended using country-specific pools in the 
 +''server'' commands: "you get a bit better result if you use the continental 
 +zones ... and even better time if you use the country zone". This is no longer 
 +true. They now recommend looking up the global pool ''pool.ntp.org''stating 
 +that the global pool "will usually return IP addresses for servers in or close 
 +to your country ... for most users this will give the best results"
 +It's not safe to trust specific individual clocks in the public NTP pool. This 
 +is why NTP looks at multiple clocks and compares them before it selects a 
 +clock to synchronize with. It's important to configure the clock selection 
 +process. Current best practice is to wait until at 3 of 4 public clocks 
 +contacted agree about what time it is. Add the command: 
 +  tos minclock 4 minsane 3 
 +It's recommended to set NTP to associate with an odd number of pool clocks, 
 +equal to at least //minclock// + 2. If your chosen //minclock// is 4, you can 
 +calculate your target number of pool clocks as: 
 +| //minclock// + 2 + 1 = 7 | 
 +(You can use a larger odd number if you wish, but 7 is adequate). 
 +NTP counts every clock you declare explicitly in ''/etc/ntp.conf'', plus the 
 +pool clocks it discovers, against its //maxclock// parameter. So to come up 
 +with the right limit, take the number you just calculated, and add 1 for each 
 +explicit clock declaration you have in ''/etc/ntp.conf'', //including// your 
 +''pool'' command, and use that number to set //maxclock//. For example, if you 
 +just have the one ''pool'' command and no other clocks declared, then 
 +| //maxclock// = 7 + 1 | 
 +and you should add the command: 
 +  tos maxclock 8 
 +You can easily double-check your clock associations using the command 
 +  # ntpq -n -p 
 +and verify that the number of pool clocks is what you expected. 
 +<note important> 
 +The risk that you run if you don't set //minclock//, //minsane//, and 
 +//maxclock// properly is that the NTP clock selection algorithm will get it 
 +wrong at boot time and give you inaccurate time, or even panic and exit. 
 +===== Correcting for a fast or slow hardware clock ===== 
 +Any hardware clock runs a few parts per million too fast or too slow. Over 
 +time, NTP automatically calculates what this error is and compensates for it. 
 +It can also store its calculation in a file that it re-reads when restarted. 
 +The preinstalled Slackware ''/etc/ntp.conf'' already has the necessary 
 +command in it to enable this feature: 
 +  driftfile /var/lib/ntp/drift 
 +===== Operating NTP as a server ===== 
 +Beyond access control, there is no configuration needed to let your NTP host 
 +operate as a server and supply time to your other devices. 
 +In fact, it is a good idea to make one host on your network the primary time 
 +server, and configure your other devices to get time from it. This reduces 
 +bandwidth on your uplink. Plus it reduces the load on the public NTP pool if 
 +you are using it. 
 +If you have client devices that are Slackware hosts, they should not use the 
 +''pool'' command. Intead they can use the ''server'' command and identify your 
 +primary local time host by IP address. Otherwise, they are configured much 
 +like your primary time host. 
 +You might want your other devices to stay synchronized with your primary time 
 +host even when your uplink goes down. The way this used to work was by adding 
 +your own hardware clock as a sort of "emergency" reference clock that will 
 +keep your devices in synch with one another even without an uplink. This is 
 +the approach taken by the preinstalled Slackware ''/etc/ntp.conf'': 
 +<note important> 
 +  server 
 +  fudge stratum 10 
 +Use of this clock driver is no longer recommended. 
 +The local clock driver has been replaced by //Orphan Mode//. The commands 
 +above should be changed to: 
 +  tos orphan 10 
 +Orphan mode is also useful in a closed environment, say a high-security 
 +installation, where you will not use the NTP pool or an actual UTC hardware 
 +clock and only want the devices on a network to agree on the time. It's less 
 +less helpful when you want a timely reminder to watch //Star Trek: Picard//. 
 +If you do run in a closed environment, NTP will have no way to calculate drift 
 +of your hardware clock against a true clock. You can make your own 
 +observations of how fast or slow your hardware clock tends to drift, and 
 +manually compensate using the command: 
 +  tinker freq NNN 
 +where //NNN// is the observed frequency error of your hardware clock in parts 
 +per million. This is mutually exclusive with the ''driftfile'' command so you 
 +will also have to take that command out. 
 +===== Startup ===== 
 +At one time it was standard practice to use the ''ntpdate'' command to make an 
 +quick rough adjustment to the system clock and then start ''ntpd'', but in 
 +this version of NTP that's no longer recommended. It is recommended instead to 
 +start ''ntpd'' as early as possible in the boot sequence and use the ''-g'' 
 +option to set the time. 
 +This is already what Slackware 15.0 is preconfigured to do. To have NTP run at 
 +startup, make ''/etc/rc.d/rc.ntpd'' an executable script: 
 +  # chmod 755 /etc/rc.d/rc.ntpd 
 +and then either reboot or start it manually: 
 +  # /etc/rc.d/rc.ntpd start 
 +The script ''/etc/rc.d/rc.ntpd'' looks for the file ''/run/ntpd.pid'' when it 
 +is passed the ''stop'' or ''status'' option. The command to create this file 
 +is already present in ''/etc/ntp.conf'': 
 +  pidfile /var/run/ntpd.pid 
 +You may have noticed that the pathnames disagree, and that's mildly 
 +infuriating but makes no actual difference because ''/var/run'' is a symbolic 
 +link to ''/run''
 +System services that should wait to start until the clock is stable can be 
 +preceded by the ''ntp-wait'' command, for example databases. For example, 
 +you could conceivably edit the MariaDB section of ''/etc/rc.d/rc.M'' to read: 
 +  # Start the MariaDB database: 
 +  if [ -x /etc/rc.d/rc.mysqld ]; then 
 +    /usr/sbin/ntp-wait -v 
 +    /etc/rc.d/rc.mysqld start 
 +  fi 
 +===== Monitoring NTP ===== 
 +You can review how the pool discovery process is working with the command 
 +  # ntpq -n -p 
 +Here is some sample output:
 <code> <code>
-# ntpq -p 
      remote           refid      st t when poll reach   delay   offset  jitter      remote           refid      st t when poll reach   delay   offset  jitter
 ============================================================================== ==============================================================================
-*     3 u  916 1024  377    0.367    7.897   2.552 + pool.ntp.org    .POOL.          16 p    -   64    0    0.000   +0.000   0.000 
- LOCAL(0)        .LOCL         10 l  10h   64    0    0.000    0.000   0.000+-    u  665 1024  377   20.524   -1.211   0.030 
 +-   .SOCK.           1 u 1000 1024  377   51.379   -4.051   0.166 
 ++      3 u  966 1024  377    1.644   +0.731   0.197 
 +*      u  748 1024  377   40.568   +0.372   0.119 
 +-       2 u  905 1024  377   16.545   -1.035   0.147 
 ++      3 u  556 1024  377    1.565   +0.827   0.076 
 ++       2 u  220 1024  377    0.325   -1.269   0.865
 </code> </code>
-<note tip>Againyou might have to wait a few minutes before the first synchronization takes place.</note> +In a nutshellthis tells us we have 1 master pool source (type="p") plus 7 
-   +peers (type "u")It also tells us:
-===== Monitor the performance of ntpd =====+
-You will notice that the logfile ''/var/log/ntp.log'' does not contain any information about the actual accuracy of your system clockIf it's important to you, you can log the statistics of time corrections applied by NTP daemon to the system clockTo do this, add the following lines to ''/etc/ntp.conf'':+  * Our system peer (that we are getting our time from) is ("*"). 
 +  * We have three more peer candidates ("+") in case our system peer goes away. 
 +  * NTP is considering pruning three outliers ("-").
-<code> +The ''loopstats'' file and other statistics files are described in detail at 
-statsdir /var/log/ntp/ +"[[https://doc.ntp.org/documentation/4.2.8-series/monopt/|Monitoring 
-statistics loopstats +Options]]" at //[[http://doc.ntp.org|The NTP Project]]//. Here's a sample and 
-filegen loopstats file loops type day link enable +a brief explanation:
-You have to create the statsdir manuallyOnce the configuration changes are in effect, ''ntpd'' will create files named ''loops.YYYYMMDD'' in that directoryBelow is an example line from one of these files: +<file> 
-<code> +59760 66558.710 0.000038135 4.882 0.000265925 0.027674 10 
-56690 3950.569 0.001199636 2.297 0.001830770 0.571576 10 +59760 68002.750 0.000538317 4.885 0.000305204 0.025906 10 
-</code>+59760 69002.727 0.000199760 4.885 0.000309570 0.024235 10 
 +59760 69437.711 0.000656689 4.886 0.000331590 0.022673 10 
 +59760 71213.750 0.000522794 4.890 0.000313766 0.021244 10 
 +59760 73341.750 0.000484582 4.951 0.000293812 0.029264 10 
 +59760 75485.750 0.000486984 5.011 0.000274837 0.034780 10 
 +59760 76374.727 -0.000069057 5.011 0.000323638 0.032534 10 
 +59760 77579.750 0.000202697 5.012 0.000317617 0.030434 10 
 +59760 79162.710 -0.000183575 5.011 0.000326988 0.028471 10 
-The first and second number are the UTC time (expressed as [[http://en.wikipedia.org/wiki/Julian_day|Modified Julian Date]] and seconds elapsed since midnight). The third and fourth number are the offsets of time (in seconds) and of frequency (in parts per million). The fifth and sixth number are their respective uncertainties.+^ Column 1 | Modified Julian Date of the observation | 
 +^ Column 2 | Time since midnight (seconds) |
-To monitor the performance of ''ntpd'', you can examine the plot of clock offset or frequency offset vstime:+You can combine and convert these to a UNIX 
 +system value for plotting, as shown below.
-<code> +^ Column 3 | Difference observed between your system clock and your time source (seconds) | 
-$ awk '{printf("%f %f %f\n", $1+$2/86400, $3, $5)}' /var/log/ntp/loops.* > time +^ Column 5 | Column 3's jitter | 
-$ awk '{printf("%f %f %f\n", $1+$2/86400, $4, $6)}' /var/log/ntp/loops.* > freq + 
-gnuplot +^ Column 4 | Difference between your system clock frequency and the time source frequency (parts per million
-gnuplot> set xzeroaxis +^ Column 6 | Column 4's jitter | 
-gnuplot> plot 'time' with yerror + 
-gnuplot> plot 'freq' with yerror +Here is a sample **gnuplot** program that charts recent loop statistics. 
 +<file gnuplot> 
 +#! /usr/bin/gnuplot -ps 
 +# Input - four most recent loopstats files 
 +filelist=system("ls -rt /var/lib/ntp/stats/loopstats.* | tail -4") 
 +# Output - X server 
 +set terminal x11 
 +# Multiplot layout 
 +set multiplot layout 2,1 
 +# Settings common to both graphs 
 +set xlabel "Modified Julian Days (d)" 
 +set xdata time 
 +set xtics format "%F\n%T" 
 +set xzeroaxis linetype 1 
 +set grid 
 +set errorbars small linecolor "dark-gray" 
 +set key left bottom box 
 +# Time Offset graph 
 +set title 'local NTP clock: Time Offset' font ',20' 
 +set ylabel "Time Offset +/- RMS Jitter (ms)" 
 +plot [] [] for [filename in filelist] filename \ 
 +        using (86400.0*($1-40587)+$2):(1000.0*$3):(1000.0*$5) \ 
 +        title filename \ 
 +        with yerrorbars pointtype 1
-Given enough data, visual examination of the plots will allow you to see peculiarities in ''ntpd'' performance, should they arise. For example, in the case illustrated by the figure below, the rapid decrease of the frequency offset was caused by replacing the power supply unit of the machine.+# Frequency Offset graph 
 +set title 'local NTP clock: Frequency Offsetfont ',20' 
 +set ylabel "Frequency Offset +/- Allan deviation (PPM)" 
 +plot [] [] for [filename in filelist] filename \ 
 +        using (86400.0*($1-40587)+$2):($4):($6) \ 
 +        title filename \ 
 +        with yerrorbars pointtype 1 
-{{:howtos:network_services:ntp_time_freq.png|}}+Sample output: 
 +{{ :howtos:network_services:ntp-gnuplot-example.png |}}
 ====== Sources ====== ====== Sources ======
Line 191: Line 446:
   * Originally written by [[wiki:user:kikinovak|Niki Kovacs]]   * Originally written by [[wiki:user:kikinovak|Niki Kovacs]]
   * Performance monitoring section contributed by Dominik Drobek   * Performance monitoring section contributed by Dominik Drobek
 +  * Rewritten and updated to current best practice by [[wiki:user:metaed|Edward McGuire]].
 <!-- Please do not modify anything below, except adding new tags.--> <!-- Please do not modify anything below, except adding new tags.-->
 <!-- You must remove the tag-word "template" below before saving your new page --> <!-- You must remove the tag-word "template" below before saving your new page -->
-{{tag>howtos time clock synchronization author_kikinovak}}+{{tag>howtos time clock synchronization author_kikinovak slackware_15.0 author_metaed}}
 howtos:network_services:ntp ()