Home / server / Questions / 831900
Accepted
Kojo
Asked: 2017-02-11 11:11:37 +0800 CST

Syncronize SOA serial numbers explanation

  • 772

I made two successive mistakes while updating zone records. Once I forgot to increment the serial number, then next time I did put a digit more ie 20170210111 instead of 2017021011, and each time I reloaded config : rndc reload.

When I realised it, I switched back serial to a 10 digits number, so my zone file SOA serial is now 2017021012.

I get this now :

#  host -C domain.tld
Nameserver X.X.X.X:
        domain.tld has SOA record X.X.X.X.ovh.net. postmaster.domain.tld. 2017021010 28800 7200 1209600 3600
Nameserver Y.Y.Y.Y: 
        domain.tld has SOA record X.X.X.X.ovh.net. postmaster.domain.tld. 2017021003 28800 7200 1209600 86400

Where X.X.X.X is my primary DNS server IP and Y.Y.Y.Y secondary one.

I'm not really used to DNS configuration, and I really don't know how to resynchronize serials and permit propagation. I read already a lot of posts, I'm afraid I still don't know if I should give zone record a greater number, what happened when I reloaded with 13 digits...?

debian bind dns-zone soa-record

1 Answers

  1. Best Answer

    Since 20170210111 is outside the range of a 32 bit integer, presumably the zone failed to load with the impossible to represent SOA record (check your logs).

    First of all, verify (using eg dig @master example.com SOA +norec) that the master is now responding properly and with the current serial (2017021012). If not, rndc reload triggers a reload.

    When the master works properly, the slaves should update on their next refresh (based on the SOA REFRESH interval). Also, assuming that the slaves are notify-aware and with proper configuration, you should be able to trigger an immediate refresh using rndc notify example.com, as well as automatically when loading a new version of the zone.

    If this doesn't work I would think that your problem is something beyond having had an unloadable zone for a while.

    • 3