[LEDE-DEV] busybox ntpd not syncing if any server is bad (update, if _first_ server is bad)

Karl Palsson karlp at tweak.net.au
Fri Oct 27 09:52:43 PDT 2017

Magnus Kroken <mkroken at gmail.com> wrote:
> On 21.08.2017 16:34, Karl Palsson wrote:
> > on master, even with the ntpd patch for busybox applied.
> > # ntpdate -q localhost
> > server ::1, stratum 0, offset 0.000000, delay 0.00000
> > server, stratum 0, offset 0.000000, delay 0.00000 21
> > Aug 14:26:24 ntpdate[1392]: no server suitable for
> > synchronization found
> > # echo $?
> > 1
> > # ntpd -w -d -p localhost
> > ntpd: sending query to
> > ntpd: reply from peer is unsynced
> > ntpd: sending query to
> Thanks for the clarification. Running ntpd like before (-p
> bad.example.org -p ntp.uio.no), LEDE master with Busybox 1.27.2
> does seem to work:
> root at LEDE:~# ntpdate -q ::1
> server ::1, stratum 3, offset 0.000028, delay 0.02777
> 23 Aug 19:12:55 ntpdate[5416]: adjust time server ::1 offset
> 0.000028 sec root at LEDE:~# echo $? 0
> I also tested against my OpenWrt 15.05 VM and the current LEDE
> x86_64 snapshot (r4723), and they both give the expected output
> as you describe (OpenWrt similar to above, LEDE fails).
> Note that my patch in Patchwork only updates to Busybox 1.27.1,
> I will send the 1.27.2 patch shortly, after rebasing and
> testing it.

For a followup, on master, with busybox 1.27.2, busybox ntpd will
sync with bad servers in the list, but only if they're not
_first_ Further, "bad.example.org" won't count as "bad" for the
purpose of these tests. Someone (presumably dnsmasq?) is
responding instantly with failures there.

Try something like 
/usr/sbin/ntpd -n -N -d -l -S /usr/sbin/ntpd-hotplug -p bad1.n
otexample.notorg -p bad2.notexample.notorg -p

It appears that the dns resolution times out, and is ordered in
such a way that busybox sends a request, and even though it gets
a reply "instantly" it doesn't actually process it for several
seconds. At that point, it .. fails? to process properly, and so
just continues sending requests, getting replies, and never
stepping or setting the stratum. (And hence, my tests with
ntpdate -q never succeeded)

I've filed https://bugs.busybox.net/show_bug.cgi?id=10466 with
busybox for this. It's actually only a bother for me in my
development environment, where the *.pool.ntp.org addresses
aren't resolveable

Karl Palsson
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.html
Type: application/pgp-signature
Size: 1161 bytes
Desc: OpenPGP Digital Signature
URL: <http://lists.infradead.org/pipermail/lede-dev/attachments/20171027/018b8afc/attachment.sig>

More information about the Lede-dev mailing list