[Dnsmasq-discuss] ip address not released
Hello, we have to repair some embedded systems with mac addresses within a specific range. These system needs bootp to get flashed some special software. So i configured the following entries (Dnsmasq version 2.40): dhcp-host=00:9a:7e:8f:*:*,net:embeddeds,embed,10m Imho the device, who fits to the mac range, will receive the address for embed with a lease time of 10 minutes Also i have a bootp entry: dhcp-boot=net:embeddeds,embed_factory.flash,embed,192.168.1.83 Everything works well with one system but the address won't be released after 10 minutes. When i replace the system with another one from the mac range i get the following error: dnsmasq[21376]: BOOTP(eth23) 192.168.1.119 00:9a:7e:8f:19:5a address in use I always have to remove the lease entry manually from /var/lib/misc/dnsmasq.leases and restart dnsmasq to get the new system running, independently from how long i wait. peter
Re: [Dnsmasq-discuss] ip address not released
Peter Katzmann wrote: Hello, we have to repair some embedded systems with mac addresses within a specific range. These system needs bootp to get flashed some special software. So i configured the following entries (Dnsmasq version 2.40): dhcp-host=00:9a:7e:8f:*:*,net:embeddeds,embed,10m Imho the device, who fits to the mac range, will receive the address for embed with a lease time of 10 minutes Also i have a bootp entry: dhcp-boot=net:embeddeds,embed_factory.flash,embed,192.168.1.83 Everything works well with one system but the address won't be released after 10 minutes. When i replace the system with another one from the mac range i get the following error: dnsmasq[21376]: BOOTP(eth23) 192.168.1.119 00:9a:7e:8f:19:5a address in use I always have to remove the lease entry manually from /var/lib/misc/dnsmasq.leases and restart dnsmasq to get the new system running, independently from how long i wait. This is a limitation of BOOTP: it doesn't support the concept of a time-limited lease, so once an address is allocated, it's allocated forever: That has to be the case, since the client cannot renew a lease, it just assumes that it has use of the address forever. I'm slightly confused about what is happening here: to get the behaviour you describe, you must be allocating the address 192.168.1.119 to the MAC address somewhere - otherwise dnsmasq will complain that there's no allocated address. Either there's an IP address in you dhcp-host line, or something in /etc/ethers. One workaround, which is counter-intuitive, is to stop nailing down an address for you box, and set --bootp-dynamic. That way each new box will get a new address. Eventually, you will have a load of addresses with infinite leases on them, which you will have to clear out, but at least you won't have to do it as often. There's actually no reason why the behaviour of dnsmasq couldn't be changed to honour the lease time in a dhcp-host statement for BOOTP: it's reasonable to assume that if it's explicitly configured, the user knows what is wanted. The default lease time associated with the dhcp-range should still be ignored, of course. I'll make the change in the next release. Cheers, Simon.
Re: [Dnsmasq-discuss] IPv6 issue
Jima on 10/06/08 14:31, wrote: On Fri, 6 Jun 2008, Adam Hardy wrote: My machine's IPv6 config seems to be up the creek. ... adam@isengard:~$ sudo ping6 ::1 Password: PING ::1(::1) 56 data bytes ping: sendmsg: Operation not permitted ping: sendmsg: Operation not permitted ping: sendmsg: Operation not permitted ping: sendmsg: Operation not permitted Uhhh. That doesn't look remotely DNS-related. To be absolutely sure, though, try using the -n flag, which disables (reverse) DNS lookups. # ping6 -n ::1 That looks like firewalling is preventing the packets from being sent, to be honest. I'd be looking more at something like: # ip6tables -L -n Which is totally outside the scope of this mailing list, but I try to be helpful...oh, here: # ping6 -n ::1 PING ::1(::1) 56 data bytes 64 bytes from ::1: icmp_seq=1 ttl=64 time=0.043 ms ^C # ip6tables -A OUTPUT -j REJECT # ping6 -n ::1 PING ::1(::1) 56 data bytes ping: sendmsg: Operation not permitted ^C That's my theory. Good theory - I hadn't realised that there is ip6tables. IPv6 is a big blind spot for me - in all my playing around with iptables and searching for a solution, I never once registered that I need ip6tables - although I see it now under the iptables man page SEE ALSO. And I see my machine has all chains set to DROP, so you're right. Thanks alot. Regards Adam
Re: [Dnsmasq-discuss] NAPTR support planned?
Johan Bergquist XR wrote: Hello everyone, Does anyone know if support for NAPTR queries (that is, being able to define NAPTR records in the same way that you can define for instance SRC and PTR records today) is planned for dnsmasq and, if so, how far away that support is? Boy, RFC2915 looks complex! However, as far as I can see, all the complexity is loaded on the client/resolver. The server just has to spit out data. Adding support to dnsmasq is simply a question of adding option-parsing code and something to add_resource_record() in rfc1035.c. The only difficulty is deciding if the global utility is owrth the extra code-size: is NAPTR support likely to be useful to a reasonable number of people? Cheers, Simon.
[Dnsmasq-discuss] Problem with multiple Local Domains
Hi, This is my first email to this list, so please be gentle! :) I've just upgraded my Ubuntu Server from 7.04 to 8.04 and consequently DNSMasq has updated to version 2.41-2Ubuntu1 acording to Synaptic Package manager. In my config file I had a line with multiple local domains - e.g.: local=/domain.name.one/ /domain.name.two/ /localhost/ When the upgrade tried to restart the updated DNSMasq, DNSMasq failed to restart. I have tried editing this line and it seems that if there is one domain listed it works fine, but with multiple domains listed, DNSMasq won't start. Is this a fault with my config, or has this feature become broken? Thanks in advance. Regards Stephen Roberts web1-nos...@hst.me.uk (do not de-mung the email address - it's valid!)