monnier
14-08-2004, 01:29
If anybody wants to run an NTP server on their WL-500g, I have compiled
both the tiny openntpd (52KB) and the classic full blown ntpd (280KB).
The clock on my box appears to be about as bad as it gets, so the
ntpclient (built-in at /usr/sbin/ntpclient) seems to be unable to "lock"
if you try to run it with the -l flag.
openntpd painfully manages to keep the clock offset within -0.1s and +0.3s
at the cost of contacting the upstream server every other minute.
ntpd fares a bit better, syncing to within -0.1s or +0.1s while only contacting the server occasionally (not enough testing time available yet to evaluate exactly how well it works, tho). The ntp.drift file says the clock is off by about 250ppm, which is rather very large (larger than what the ntpclient claims to be able to manage).
both the tiny openntpd (52KB) and the classic full blown ntpd (280KB).
The clock on my box appears to be about as bad as it gets, so the
ntpclient (built-in at /usr/sbin/ntpclient) seems to be unable to "lock"
if you try to run it with the -l flag.
openntpd painfully manages to keep the clock offset within -0.1s and +0.3s
at the cost of contacting the upstream server every other minute.
ntpd fares a bit better, syncing to within -0.1s or +0.1s while only contacting the server occasionally (not enough testing time available yet to evaluate exactly how well it works, tho). The ntp.drift file says the clock is off by about 250ppm, which is rather very large (larger than what the ntpclient claims to be able to manage).