|
|
DNS Experts,
We recently changed our DHCP lease from the default eight days to one day to
deal with a lack of IP addresses in a few subnets. This has caused our
current default DNS scavenging settings (seven, seven and seven) to be
inadequate and we are seeing multiple host records created for the same IP.
Sure, they will clean out eventually, but we have a lot of duplicates in the
mean time. I've done some research and haven't been able to find much in the
way of best practices for a DHCP lease period that is this low.
I want to scavenge as frequently as possible, but I want to be careful to
not delete any cluster names or valid host names by accident. From research,
it sounds like systems that are live on the network should be refreshing
their DNS records at least once every 24 hours.
After some consideration, I'm thinking of setting my no-refresh interval to
one day, refresh interval to two days, and then configure one DNS server to
scavenge at least once per day (maybe more). I would be doing the scavenging
on a DNS server that is only lightly used, so I'm not concerned about
overloading it with frequent scavenging. With the configuration above, a
stale DNS record should never survive more than four days.
We have about 2,000 workstations/laptops and roughly 300 servers. We have
seven sites with large WAN pipes between them so the increase in replication
traffic is not a huge concern.
Does anyone see any problems with the settings that I'm proposing above?
Ted
|
|
The problem I see is that you are treating the symptom and ignoring the
ailment. The "solution" is in a more methodical rethinking of you IP scheme,
an overhaul of your DHCP scope (evaluate the possibility of using
super-scope) with a view towards making sufficient IPs available for your
needs.
You are using private IPs, right?
What you have described will give you a lot of pains, although you may be
able to get by.
Deji
"Tedster" <Tedster[ at ]discussions.microsoft.com> wrote in message
news:61C0E1D3-8D41-4FD2-9488-A335D2FB0DBD[ at ]microsoft.com...
[Quoted Text] > DNS Experts,
>
> We recently changed our DHCP lease from the default eight days to one day
> to
> deal with a lack of IP addresses in a few subnets. This has caused our
> current default DNS scavenging settings (seven, seven and seven) to be
> inadequate and we are seeing multiple host records created for the same
> IP.
> Sure, they will clean out eventually, but we have a lot of duplicates in
> the
> mean time. I've done some research and haven't been able to find much in
> the
> way of best practices for a DHCP lease period that is this low.
>
> I want to scavenge as frequently as possible, but I want to be careful to
> not delete any cluster names or valid host names by accident. From
> research,
> it sounds like systems that are live on the network should be refreshing
> their DNS records at least once every 24 hours.
>
> After some consideration, I'm thinking of setting my no-refresh interval
> to
> one day, refresh interval to two days, and then configure one DNS server
> to
> scavenge at least once per day (maybe more). I would be doing the
> scavenging
> on a DNS server that is only lightly used, so I'm not concerned about
> overloading it with frequent scavenging. With the configuration above, a
> stale DNS record should never survive more than four days.
>
> We have about 2,000 workstations/laptops and roughly 300 servers. We have
> seven sites with large WAN pipes between them so the increase in
> replication
> traffic is not a huge concern.
>
> Does anyone see any problems with the settings that I'm proposing above?
>
>
>
> Ted
|
|
I agree i think you would be much better off by rethinking your IP strategy.
Perhaps changing to a 255.255.0.0 subnet mask would give you literally
thousands more ip addresses to play with (assuming you are using
255.255.255.0 at the moment) Also if your network is spread over multiple
sites then you should be splitting it into multiple subnets perhaps with a
DHCP server in each one or as already suggested use superscopes on your
central DHCP server.
James.
--
James Yeomans, BSc, MCSE
Ask me directly at: http://www.justaskjames.co.uk
"A, Deji" wrote:
[Quoted Text] > The problem I see is that you are treating the symptom and ignoring the
> ailment. The "solution" is in a more methodical rethinking of you IP scheme,
> an overhaul of your DHCP scope (evaluate the possibility of using
> super-scope) with a view towards making sufficient IPs available for your
> needs.
>
> You are using private IPs, right?
>
> What you have described will give you a lot of pains, although you may be
> able to get by.
>
> Deji
>
> "Tedster" <Tedster[ at ]discussions.microsoft.com> wrote in message
> news:61C0E1D3-8D41-4FD2-9488-A335D2FB0DBD[ at ]microsoft.com...
> > DNS Experts,
> >
> > We recently changed our DHCP lease from the default eight days to one day
> > to
> > deal with a lack of IP addresses in a few subnets. This has caused our
> > current default DNS scavenging settings (seven, seven and seven) to be
> > inadequate and we are seeing multiple host records created for the same
> > IP.
> > Sure, they will clean out eventually, but we have a lot of duplicates in
> > the
> > mean time. I've done some research and haven't been able to find much in
> > the
> > way of best practices for a DHCP lease period that is this low.
> >
> > I want to scavenge as frequently as possible, but I want to be careful to
> > not delete any cluster names or valid host names by accident. From
> > research,
> > it sounds like systems that are live on the network should be refreshing
> > their DNS records at least once every 24 hours.
> >
> > After some consideration, I'm thinking of setting my no-refresh interval
> > to
> > one day, refresh interval to two days, and then configure one DNS server
> > to
> > scavenge at least once per day (maybe more). I would be doing the
> > scavenging
> > on a DNS server that is only lightly used, so I'm not concerned about
> > overloading it with frequent scavenging. With the configuration above, a
> > stale DNS record should never survive more than four days.
> >
> > We have about 2,000 workstations/laptops and roughly 300 servers. We have
> > seven sites with large WAN pipes between them so the increase in
> > replication
> > traffic is not a huge concern.
> >
> > Does anyone see any problems with the settings that I'm proposing above?
> >
> >
> >
> > Ted
>
>
|
|
Thank you for the replies and I agree with your conclusions and this was the
direction I tried to go before shortening the DHCP scopes lease. However,
the network and IP pools are controlled by a different group and they can't
or won't increase them at this time.
So back to my original question -- are the scavenging settings I proposed
workable or will they potentially cause problems?
|
|
I think what you are doing will work but is certainly not the best way to go
about things.
James.
--
James Yeomans, BSc, MCSE
Ask me directly at: http://www.justaskjames.co.uk
"Tedster" wrote:
[Quoted Text] > Thank you for the replies and I agree with your conclusions and this was the
> direction I tried to go before shortening the DHCP scopes lease. However,
> the network and IP pools are controlled by a different group and they can't
> or won't increase them at this time.
>
> So back to my original question -- are the scavenging settings I proposed
> workable or will they potentially cause problems?
|
|
James,
Thank you for taking the time to reply.
Ted
"James Yeomans BSc, MCSE" wrote:
[Quoted Text] > I think what you are doing will work but is certainly not the best way to go > about things. > James. > -- > James Yeomans, BSc, MCSE > Ask me directly at: http://www.justaskjames.co.uk> > > "Tedster" wrote: > > > Thank you for the replies and I agree with your conclusions and this was the > > direction I tried to go before shortening the DHCP scopes lease. However, > > the network and IP pools are controlled by a different group and they can't > > or won't increase them at this time. > > > > So back to my original question -- are the scavenging settings I proposed > > workable or will they potentially cause problems? |
|
|