import of dnsmasq-2.9.tar.gz

[people/ms/dnsmasq.git] / FAQ

Q: Why does dnsmasq open UDP ports >1024 as well as port 53.
   Is this a security problem/trojan/backdoor?

A: The high ports that dnsmasq opens is for replies from the upstream
   nameserver(s). Queries from dnsmasq to upstream nameservers are sent
   from these ports and replies received to them. The reason for doing this is
   that most firewall setups block incoming packets _to_ port 53, in order
   to stop DNS queries from the outside world. If dnsmasq sent its queries
   from port 53 the replies would be _to_ port 53 and get blocked.

   This is not a security hole since dnsmasq will only accept replies to that
   port: queries are  dropped. The replies must be to oustanding queries
   which dnsmasq has forwarded, otherwise they are dropped too.
 
   Addendum: dnsmasq now has the option "query-port" (-Q), which allows
   you to specify the UDP port to be used for this purpose.  If not
   specified, the operating system will select an available port number
   just as it did before.
 
Q: Why doesn't dnsmasq support DNS queries over TCP? Don't the RFC's specify
   that?

A: Yes, they do, so technically dnsmasq is not RFC-compliant. In practice, the
   sorts of queries which dnsmasq is used for are always sent via UDP. Adding 
   TCP support would make dnsmasq much more heavyweight for no practical 
   benefit. If you really want to do zone transfers, forward port 53 TCP 
   using in-kernel port-forwarding or a port-fowarder like rinetd.


Q: When I send SIGUSR1 to dump the contents of the cache, some entries have
   no IP address and are for names like mymachine.mydomain.com.mydomain.com.
   What are these?

A: They are negative entries: that's what the N flag means. Dnsmasq asked 
   an upstream nameserver to resolve that address and it replied "doesn't 
   exist, and won't exist for <n> hours" so dnsmasq saved that information so
   that if _it_ gets asked the same question it can answer directly without
   having to go back to the upstream server again. The strange repeated domains
   result from the way resolvers search short names. See "man resolv.conf" for
   details.


Q: Will dnsmasq compile/run on non-Linux systems?

A: Yes, there is explicit support for *BSD and Solaris.
   For other systems, try altering the settings in config.h.

A: Update for V2. Doing DHCP is rather non-portable, so there may be
   a few teething troubles. The initial 2.0 release is known to work
   on Linux 2.2.x, Linux 2.4.x and Linux 2.6.x with uclibc and glibc
   2.3. It also works on FreeBSD 4.8. The crucial problem is sending
   raw packets, bypassing the IP stack. Dnsmasq contains code to do
   using PF_PACKET sockets (which is for Linux) and the Berkeley packet
   filter (which works with BSD). If you are trying to port to another
   Un*x, bpf is the most likeley candidate. See config.h 

Q: My companies' nameserver knows about some names which aren't in the
   public DNS. Even though I put it first in /etc/resolv.conf, it
   dosen't work: dnsmasq seems not to use the nameservers in the order
   given. What am I doing wrong?

A: By default, dnsmasq treats all the nameservers it knows about as
   equal: it picks the one to use using an algorithm designed to avoid 
   nameservers which aren't responding. To make dnsmasq use the
   servers in order, give it the -o flag. If you want some queries
   sent to a special server, think about using the -S flag to give the
   IP address of that server, and telling dnsmasq exactly which
   domains to use the server for.

Q: OK, I've got queries to a private nameserver working, now how about 
   reverse queries for a range of IP addresses?

A: Use the standard DNS convention of <reversed address>.in-addr.arpa.
   For instance to send reverse queries on the range 192.168.0.0 to 
   192.168.0.255 to a nameserver at 10.0.0.1 do
   server=/0.168.192.in-addr.arpa/10.0.0.1

Q: Dnsmasq fails to start with an error like this: "dnsmasq: bind
   failed: Cannot assign requested address". What's the problem?

A: This has been seen when a system is bringing up a PPP interface at
   boot time: by the time dnsmasq start the interface has been
   created, but not brought up and assigned an address. The easiest
   solution is to use --interface flags to specify which interfaces
   dnsmasq should listen on. Since you are unlikely to want dnsmasq to
   listen on a PPP interface and offer DNS service to the world, the
   problem is solved.

Q: I'm running on BSD and dnsmasq won't accept long options on the
   command line. 

A: Dnsmasq when built on BSD systems doesn't use GNU getopt by
   default. You can either just use the single-letter options or
   change config.h and the Makefile to use getopt-long. Note that
   options in /etc/dnsmasq.conf must always be the long form,
   on all platforms.

Q: Names on the internet are working fine, but looking up local names 
   from /etc/hosts or DHCP doesn't seem to work.

A: Resolver code sometime does strange things when given names without
   any dots in. Win2k and WinXP may not use the DNS at all and just
   try and look up the name using WINS. On unix look at "options ndots:"
   in "man resolv.conf" for details on this topic. Testing lookups
   using "nslookup" or "dig" will work, but then attempting to run
   "ping" will get a lookup failure, appending a dot to the end of the
   hostname  will fix things. (ie "ping myhost" fails, but "ping
   myhost." works. The solution is to make sure that all your hosts
   have a domain set ("domain" in resolv.conf, the network applet in
   windows, or set a domain in your DHCP server). Any domain  will do,
   but "localnet" is traditional. Now when you resolve "myhost" the
   resolver will attempt to look up "myhost.localnet" so you need to
   have dnsmasq reply to that name. The way to do that is to include
   the domain in each name on /etc/hosts and/or to use the
   --expand-hosts and --domain-suffix options.

Q: Can I get dnsmasq to save the contents of its cache to disk when
   I shut my machine down and re-load when it starts again.

A: No, that facility is not provided. Very few names in the DNS have
   their time-to-live set for longer than a few hours so most of the
   cache entries would have expired after a shutdown. For longer-lived
   names it's much cheaper to just reload them from the upstream
   server. Note that dnsmasq is not shut down between PPP sessions so
   go off-line and then on-line again will not lose the contents of
   the cache.

Q: Who are Verisign, what do they have to do with the bogus-nxdomain
   option in dnsmasq and why should I wory about it?

A: [note: this was written in September 2003, things may well change.]
   Versign run the .com and .net top-level-domains. They have just
   changed the configuration of their servers so that unknown .com and
   .net domains, instead of returning an error code NXDOMAIN, (no such
   domain) return the address of a host at Versign which runs a web
   server showing a search page. Most right-thinking people regard
   this new behaviour as broken :-).  You can test to see if you are
   suffering Versign brokeness by run a command like 
   
   host jlsdajkdalld.com

   If you get "jlsdajkdalld.com" does not exist, then all is fine, if
   host returns an IP address, then the DNS is broken. (Try a few
   different unlikely domains, just in case you picked a wierd one
   which really _is_ registered.)

   Assuming that your DNS is broken, and you want to fix it, simply
   note the IP address being returned and pass it to dnsmasq using the
   --bogus-nxdomain flag. Dnsmasq will check for results returning
   that address and substitute an NXDOMAIN instead. 

   As of writing, the IP address in question for the .com and .net
   domains is is 64.94.110.11. Various other, less prominent,
   registries pull the same stunt; there is a list of them all, and
   the addresses to block, at http://winware.org/bogus-domains.txt

Q: This new DHCP server is well and good, but it doesn't work for me.
   What's the problem?

A: There are a couple of configuration gotchas which have been
   encountered by people moving from the ISC dhcpd to the dnsmasq
   integrated DHCP daemon. Both are related to differences in 
   in the way the two daemons bypass the IP stack to do "ground up"
   IP configuration and can lead to the dnsmasq daemon failing
   whilst the ISC one works.

   The first thing to check is the broadcast address set for the
   ethernet interface. This is normally the adddress on the connected
   network with all ones in the host part. For instance if the 
   address of the ethernet interface is 192.168.55.7 and the netmask
   is 255.255.255.0 then the broadcast address should be
   192.168.55.255. Having a broadcast address which is not on the
   network to which the interface is connected kills things stone
   dead.

   The second potential problem relates to firewall rules: since the ISC 
   daemon in some configurations bypasses the kernel firewall rules 
   entirely, the ability to run the ISC daemon does not indicate 
   that the current configuration is OK for the dnsmasq daemon.
   For the dnsmasq daemon to operate it's vital that UDP packets to 
   and from ports 67 and 68 and broadcast packets with source 
   address 0.0.0.0 and destination address 255.255.255.255 are not 
   dropped by iptables/ipchains.

Q: I'm running Debian, and my machines get an address fine with DHCP,
   but their names are not appearing in the DNS.

A: By default, none of the DHCP clients send the host-name when asking
   for a lease. For most of the clients, you can set the host-name to
   send with the "hostname" keyword in /etc/network/interfaces. (See
   "man interfaces" for details.) That doesn't work for dhclient, were
   you have to add something like "send host-name daisy" to
   /etc/dhclient.conf

Q: I'm network booting my machines, and trying to give them static
   DHCP-assigned addresses. The machine gets its correct address
   whilst booting, but then the OS starts and it seems to get
   allocated a different address.

A: What is happening is this: The boot process sends a DHCP
   request and gets allocated the static address corresponding to its
   MAC address. The boot loader does not send a client-id. Then the OS
   starts and repeats the DHCP process, but it it does send a
   client-id. Dnsmasq cannot assume that the two requests are from the
   same machine (since the client ID's don't match) and even though
   the MAC address has a static allocation, that address is still in
   use by the first incarnation of the machine (the one from the boot,
   without a client ID.) dnsmasq therefore has to give the machine a
   dynamic address from its pool. There are three ways to solve this:
   (1) persuade your DHCP client not to send a client ID, or (2) set up
   the static assignment to the client ID, not the MAC address. The
   default client-id will be 01:<MAC address>, so change the dhcp-host
   line from "dhcp-host=11:22:33:44:55:66,1.2.3.4" to
   "dhcp-host=id:01:11:22:33:44:55:66,1.2.3.4" or (3) tell dnsmasq to
   ignore client IDs for a particular MAC address, like this:
   dhcp-host=11:22:33:44:55:66,id:*

Q: What network types are supported by the DHCP server?
  
A: Ethernet (and 802.11 wireless) are supported on all platforms. On
   Linux Token Ring is also supported.

Q: What is this strange "bind-interface" option?

A: The DNS spec says that the reply to a DNS query must come from the
   same address it was sent to. The traditional way to write an UDP
   server to do this is to find all of the addresses belonging to the
   machine (ie all the interfaces on the machine) and then create a
   socket for each interface which is bound to the address of the
   interface. Then when a packet is sent to address A, it is received
   on the socket bound to address A and when the reply is also sent
   via that socket, the source address is set to A by the kernel and
   everything works. This is the how dnsmasq works when
   "bind-interfaces" is set, with the obvious extension that is misses
   out creating sockets for some interfaces depending on the
   --interface, --address and --except-interface flags.  The
   disadvantage of this approach is that it breaks if interfaces don't
   exist or are not configured when the daemon starts and does the
   socket creation step. In a hotplug-aware world this is a real
   problem.

   The alternative approach is to have only one socket, which is bound
   to the correct port and the wildcard IP address (0.0.0.0). That
   socket will receive _all_ packets sent to port 53, no matter what
   destination address they have. This solves the problem of
   interfaces which are created or reconfigured after daemon
   start-up. To make this work is more complicated because of the
   "reply source address" problem. When a UDP packet is sent by a
   socket bound to 0.0.0.0 its source address will be set to the
   address of one of the machine's interfaces, but which one is not
   determined and can vary depending on the OS being run. To get round
   this it is neccessary to use a scary advanced API to determine the
   address to which a query was sent, and force that to be the source
   address in the reply. For IPv4 this stuff in non-portable and quite
   often not even available (It's different between FreeBSD 5.x and
   Linux, for instance, and FreeBSD 4.x, Linux 2.0.x and OpenBSD don't
   have it at all.) Hence "bind-interfaces" has to always be available
   as a fall back. For IPv6 the API is standard and universally
   available.

   It could be argued that if the --interface or --address flags are
   used then binding interfaces is more appropriate, but using
   wildcard binding means that dnsmasq will quite happily start up
   after being told to use interfaces which don't exist, but which are
   created later. Wildcard binding breaks the scenario when dnsmasq is
   listening on one interface and another server (most probably BIND)
   is listening on another. It's not possible for BIND to bind to an
   (address,port) pair when dnsmasq has bound (wildcard,port), hence
   the ability to explicitly turn off wildcard binding.