Using named_dump.db to preserve cache across reboots

Mark Damrose mdamrose at elgin.cc.il.us
Tue Mar 12 17:27:24 UTC 2002


"Mr Toad" <rat_love_cat at yahoo.co.uk> wrote in message
news:a6lcj9$dh at pub3.rc.vix.com...
>
> On 11 Mar 2002 09:16:28 -0800, Jim Reid <jim at rfc1035.com> wrote:
> >Why? What's the point of preserving a cache that is by definition now
> >out of date? Would you expect the weather forecast to tell you what
> >the weather was like 2 weeks ago? And if it did, what use would that
> >be to anyone?
>
> The reason for wanting to preserve the cache is because, over time, my
> nameserver builds up a rich cache which speeds up lookups for
> commonly-requested names.  Now if I reboot the machine, that cache is
> lost and will take time to grow back.  I would prefer to be able to
> preserve the cache across reboots so that it is not lost.
>
> I can't understand why you say that it's like asking for last week's
> weather forecast.  Surely the nameserver cache is just as valid after
> a reboot as it was a couple of minutes before?

I don't see the big deal.  Records in the cache expire and are looked up
again all the time.  For the popular names, they will be looked up again
quickly, then they will be in your cache.

However, more and more of the popular services use load balancers.  They
have very short TTL on their records, so that adjustments are seen quickly.
This makes your cache less usefull.  When you have a TTL of 5 minutes and a
reboot takes 2-3, it's not worth reloading that record.

If you really want this cache to be re-built, you could always write a
script to parse the named_dump.db file and lookup everything in it.

>
> Mr Toad
>
>




More information about the bind-users mailing list