home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.protocols.time.ntp
- Path: sparky!uunet!walter!porthos!iscp.bellcore.com!jhc
- From: jhc@iscp.bellcore.com (Jonathan Clark)
- Subject: Newbie xntp user seeks enlightenment
- Organization: Bellcore
- Date: Wed, 2 Sep 92 15:15:15 GMT
- Message-ID: <1992Sep2.151515.10956@porthos.cc.bellcore.com>
- Sender: netnews@porthos.cc.bellcore.com (USENET System Software)
- Lines: 106
-
- Greetings.
-
- I have XNTP v3.1 (tar file dated Aug 24 from louie) up and running on an
- IBM RS6000 workstation under AIX 3.2, and I have some naive questions
- about what I'm seeing when I query the server and in the logfile. Rather
- than hacking through the code, I thought I'd ask the net. Note that I
- am by no means a guru in this topic, and my use of language is probably
- woefully imprecise, but please bear with me.
-
- For example, I have the following output:
-
- rain$ /usr/lbin/xntpdc -p
- remote local st poll reach delay offset disp
- ======================================================================
- *flash.bellcore. 128.96.144.169 1 64 7 0.0144 0.009685 7.8921
-
- I read this as meaning that I'm:
-
- - I'm currently synchronized to flash
- - which is running at stratum 1 (it has a WWVB clock),
- - I poll flash every 64 seconds,
- - the next poll is in 37 seconds,
- - the reachability was 7 (octal)
- - the filter delay was 0.0144s, 14.4ms,
- - the offset was -0.009685s, 9.68ms, and
- - the dispersion was 7.8921s, or 7892.1ms.
-
- First set of questions. Am I correct in believing that:
-
- - the reachability is the number of queries which were
- successfully processed, so in this case the last 3 packets
- were good (and something nasty happened before that)?
- - the filter delay is measurement of how long it takes xntpd
- to do its processing?
- - the offset is the amount the server differs from its clock?
- - the dispersion is the difference between the local time and
- the time reported by the server?
-
- From the 'sysinfo' command of xntpdc I have a sync distance of 0.0115
- and a sync dispersion of 1.3065 (this was about 15 minutes later).
-
- - the sync distance is an estimate of the round-trip packet
- time in seconds?
- - the sync dispersion means the same as above, but the time
- difference between me and flash is now rather less.
-
- Consecutively to the above I issued an 'ntpq -p' command, which gave me:
-
- rain$ /usr/lbin/ntpq -p
- remote refid st when poll reach delay offset disp
- =========================================================================
- #flash.bellcore. .WWVB. 1 37 64 7 14.4 9.68 7892.1
-
- ie basically the same thing except with a factor of 1000 thrown in.
- From this I conclude that one of the two man pages is in error, since
- both claim that they print out their results in seconds. I presume that
- xntpdc actually produces seconds and ntpq milliseconds, am I right?
-
- Turning now to the log files, I think I have the various system and peer
- status flags sorted out. However, I get messages like these occasionally:
-
- offset -0.006618 freq -0.76593 comp 0
-
- This is part of the hourly statistics. The offset is as above (the
- amount the kernel clock is off?), and am I correct in thinking that
- the freq field is the amount the hardware clock needs adjusting to
- keep perfect time? comp seems to be ``log2 of time constant'', but
- what's that?
-
- adjust: STEP dropped (128.96.32.20 offset 0.556186199)
- adjust: STEP 128.96.32.20 offset 0.555446148
-
- These two mean that the difference between kernel time and the server's
- time was greater than could be corrected by slewing the clock, and that
- the kernel time was stepped? In the first case the time wasn't actually
- stepped because it was too soon since the last correction.
-
- Now, if anyone is still reading, I have some questions about my target
- configuration. This is for an isolated network without a radio clock,
- and set up pretty much as a tree. Right now I have one workstation set
- up as a master, using its local clock, with this config file:
-
- # local ref clock at stratum 1
- server 127.127.1.1
- fudge 127.127.1.1
-
- When I use xntpdc -p this successfully tells me that I'm claiming to have
- a stratum 1 clock. However, four of the five workstations I have which
- are using me as a server, all with identical binaries and config files,
- which consist of the single line:
-
- server 128.96.144.169
-
- see me as being on stratum 16, while the other one sees me at startum 2,
- which is what I expect! This one machine used to see me at stratum 16,
- but apparently changed its mind overnight. Anyone care to share some
- light about what's going on? Or can someone tell me how to set up my
- various configuration files so that this works like I want?
-
- All help and comments gratefully accepted. If there's interest and people
- respond by mail, I'll summarize.
-
- Jonathan Clark
- jhc@iscp.bellcore.com
-
- The Englishman never enjoys himself except for some noble purpose.
-