BIND8 &BIND9 Notify
Ladislav Vobr
lvobr at ies.etisalat.ae
Wed Aug 6 02:51:26 UTC 2003
not easy for some customers to do, but nice solution:-), thanks mark.
btw. perhaps having a forced zone tranfer as a bind9 option will help
too. I found many customers having a similiar problem.
Ladislav
Mark_Andrews at isc.org wrote:
>>I am having similar problem,
>>
>> What is the best way to synchronize the zone with bind9, in case the
>>client is the master and has misconfigured his zone serial number and
>>has something lower than slave but with a uptodate rrs. I used to force
>>the zone transfer on Bind8 with named-xfer manually -s 0 and then do ndc
>>reload, it re-read the file and set the lower serial which was on the
>>master server, and it was fine. With Bind 9, deleting or refreshing zone
>>will not help since serial is lower then slave has. Do I really have to
>>stop, delete the file and start ?
>>
>>......... changing the serial on the master side to higher than slave
>>will definitely solve it , but I recommended to the customer to stick to
>>YYYYMMDDXX format which was lower than the slave had.
>>
>>Ladislav
>>
>>
>
> Just rollover the serial number.
>
> Add 0x7fffffff (2147483647) to the current serial number.
>
> Wait for the slave(s) to catch up.
>
> If the number is less than 4150564249 add another 0x7fffffff
> (2147483647) and wait for the slave(s) to catchup.
>
> Make the serial 2003080600.
>
> (gdb) print (unsigned)2003080600-0x7fffffff
> $1 = 4150564249
>
> e.g.
> 2003010100 + 0x7fffffff > 4150493747
> 4150493747 + 0x7fffffff > 2003010098
> 2003010098 + 70502 -> 2003080600
>
> Mark
>
>
>
>>Mark_Andrews at isc.org wrote:
>>
>>
>>
>>>>On BIND 8.3.4 server (running as stealth server) If I delete zone file and
>>>>do a kill -HUP then the named z-xfers the zone and everything is fine
>>>>
>>>>On BIND-9.2.2
>>>>
>>>>1. If I delete the zone file then do a delete zone file, kill -HUP it does
>>>>not zone-xfer the zone file but continues to serve the data from the old
>>>>zone file.
>>>>
>>>>
>>>>
>>>>
>>> Don't do that.
>>>
>>>
>>>
>>>
>>>
>>>>2. If I do a rndc reload - same thing happens
>>>>
>>>>
>>>>
>>>>
>>> rndc reload applies to master zones.
>>>
>>>
>>>
>>>
>>>
>>>>3. If I do a rndc refresh zone IN view-name then one of two things happens
>>>>(i) if the bind9 named has only just been started then nothing happens
>>>>(ii) if the bind9 has been running for a while then the zone file gets
>>>>refreshed
>>>>
>>>>
>>>>
>>>>
>>> rndc refresh will cause named to perform a refresh query.
>>> When named starts up it automatically performs refresh
>>> queries for all zones. These are rate limited. If you
>>> issues the rndc refresh early enough you will get caught
>>> in the queue.
>>>
>>>
>>>
>>>
>>>
>>>>What is going on ??? I want to be able to, predictably, force a zone-refres
>>>>
>>>>
>>h
>>
>>
>>>>on a 'stealth' server. How do I do it ?
>>>>
>>>>DJ
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>--
>>>Mark Andrews, Internet Software Consortium
>>>1 Seymour St., Dundas Valley, NSW 2117, Australia
>>>PHONE: +61 2 9871 4742 INTERNET: Mark.Andrews at isc.org
>>>
>>>
>>>
>>>
>>>
>>
>>
>--
>Mark Andrews, Internet Software Consortium
>1 Seymour St., Dundas Valley, NSW 2117, Australia
>PHONE: +61 2 9871 4742 INTERNET: Mark.Andrews at isc.org
>
>
>
More information about the bind-users
mailing list