<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body dir="auto"><div></div><div>I'll say it, since everyone else is dancing around the issue:</div><div><br></div><div>If you are updating zone data more often than once/second and are concerned with a) encoding time into soa serial and b) have traditional secondaries doing {a,i}xfr based on serial, you're just Doing It Wrong, and most likely have other, much larger issues, and should likely consider that sending updates to BIND might not be a sustainable long-term solution.</div><div><br></div><div>If your data is in constant flux, shoehorning it into a constant series of point-in-time snapshots seems pointless and inefficient. This is probably why Route 53 doesn't bother supporting zone transfers.</div><div><br></div><div>Matt</div><div><br>On Mar 31, 2016, at 12:22 PM, Dave Warren <<a href="mailto:davew@hireahit.com">davew@hireahit.com</a>> wrote:<br><br></div><blockquote type="cite">
<meta content="text/html; charset=ISO-8859-1" http-equiv="Content-Type">
<div class="moz-cite-prefix">My point was more this: Don't update
more than once a second, any further updates can wait for the next
second.<br>
<br>
If you're using dynamic updates or something that is doing updates
in real time, fine, let it increment the serial as many times a
second as you want but if you're generating static zonefiles from
a database or updating by hand or other script where you want to
generate new serials without having to worry about some "nn" per
day counter, epoch is perfectly fine, and subsequent updates
waiting a second won't hurt you.<br>
<br>
<pre class="moz-signature" cols="72">--
Dave Warren
<a class="moz-txt-link-freetext" href="http://www.hireahit.com/">http://www.hireahit.com/</a>
<a class="moz-txt-link-freetext" href="http://ca.linkedin.com/in/davejwarren">http://ca.linkedin.com/in/davejwarren</a>
</pre>
On 2016-03-30 18:49, Roy Hooper wrote:<br>
</div>
<blockquote cite="mid:CAHW=2z7rNcZjCBru_NyaWPE_iE=s8ArdJQDjWGSn9QAXDQ2+PQ@mail.gmail.com" type="cite">
<div dir="ltr">Even with thousands of updates a day, and 86400
seconds per day, there's plenty of headroom to handle
same-second updates when they do happen...
<div><br>
</div>
<div>One approach is to use this pseudocode:</div>
<div><br>
</div>
<div>if time() > current_serial:</div>
<div> new_serial = time()</div>
<div>else:</div>
<div> new_serial = serial + 1</div>
<div><br>
</div>
<div>This handles both clock rewind scenarios (that never
happens in the real world...) as well as multiple updates in
the same second.</div>
<div>It also simply becomes a counter if you have over 86400
updates/day, although in that case the suggestion to just use
a counter makes more sense so you can start at 1.<br>
</div>
<div><br>
</div>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Wed, Mar 30, 2016 at 4:32 PM, Dave
Warren <span dir="ltr"><<a moz-do-not-send="true" href="mailto:davew@hireahit.com" target="_blank">davew@hireahit.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On 2016-03-30 12:35, Mark Andrews wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
Or you can just use it as a serial number. Encoding
date / time<br>
since epoch only works if the freqency of updates is
low.<br>
</blockquote>
<br>
</span>
Even if you have more than one update per second, isn't
incrementing the serial number and rebuilding the zone once
per second going to be sufficient for real world purposes?<br>
<br>
Or am I misunderstanding?<span class="im HOEnZb"><br>
<br>
-- <br>
Dave Warren<br>
<a moz-do-not-send="true" href="http://www.hireahit.com/" rel="noreferrer" target="_blank">http://www.hireahit.com/</a><br>
<a moz-do-not-send="true" href="http://ca.linkedin.com/in/davejwarren" rel="noreferrer" target="_blank">http://ca.linkedin.com/in/davejwarren</a><br></span></blockquote></div></div></blockquote></blockquote></body></html>