Discuss mailing list

[Jeff Taylor <shdwdrgn AT sourpuss.net>]

Actually it was the other way around...  The original OZ zone had a very large serial (probably based on epoch seconds), and it was changed to using the YYYMMDDxx format, which gave the zone a *smaller* serial.

When something like this happens, dns services do not automatically grab the new zone file.  They only get the updates when the serial has been incremented.  In cases like this, you have to wait for the zone to expire before the dns server forces the download of the new zone file.  The expire time on the oz zone is two weeks, so we would have seen failures during that entire period.  Fortunately the server operators who slave that zone were able to manually reset the file within a couple days and get everyone back in sync again.


On 06/28/2012 01:21 PM, Alex Hanselka wrote: The way a zone operates is that with every edit, the serial number in the SOA record gets incremented by one.  Usually they are in the form of something like YYYYMMDDXX where XX is the edit for that day.  The slave servers, in this case the other T1s and sometimes T2s will only update if the serial number has incremented since the one they have.  In this case, the serial for OZ got incremented to a huge amount so the other server never would have updated again causing some failures.  Also, some of the T1s had not even updated their config to resolve the OZ zone which also caused intermittent failures depending on the T1 my server queried.

On Jun 28, 2012, at 2:19 PM, Peter Green wrote:

Wow, fast reply! :-)

What do you mean when you say "serial issues"?

The reason I am interested is I am building a testing a T1.

Peter

On 28.06.2012 20:16, Alex Hanselka wrote:

Due to some T1 and serial issues, the OZ zone was not resolving

sometimes. This made me fail the test which made my "pass rate" on the

stats go down. It did not affect the overall uptime of the server however

as you can also see in the stats :)

On Jun 28, 2012, at 2:16 PM, Peter Green wrote:

Forgive me if this has been already been covered, but what were the

problems with .oz that dropped your uptime? Peter On 28.06.2012 20:11,

Alex Hanselka wrote:

My config is the simple one from the wiki. I use slave method instead

of hints. I run it on a Linode 1024 with a bunch of other stuff. I

ran it on Arch Linux until recently I have migrated to Scientific

Linux (RHEL based). My "passed" rate went down a bit this past week

with the issues with oz though, how sad. I think it is 98.98% at the

moment.

--------

You are a member of the OpenNIC Discuss list.

You may unsubscribe by emailing

discuss-unsubscribe AT lists.opennicproject.org

--------
You are a member of the OpenNIC Discuss list. You may unsubscribe by emailing discuss-unsubscribe AT lists.opennicproject.org

---

Alex Hanselka

alex AT opennicproject.org