bind 9.1.1 problem

[Jim Reid]

>>>>> "Tim" == Tim Jones <tim.jones at ntli.net> writes:

    Tim>  At peak times both times these server appear to be serving
    Tim> about 1300 queries/sec.

This is a very high query rate. It's up there in root server
territory, which is far from normal for things that are not root
servers. Why are the servers so busy? If you've got a huge number of
clients, why don't you have more name servers to shoulder the load?

    Tim>  Also can anybody confirm whether bind 8.2.3 can run on
    Tim> multiple processors under Solaris, it seems that bind 8.2.3
    Tim> can't spawn threads on to mutliple processors, is this
    Tim> correct. If not how do I make it use all processors a box has
    Tim> to offer, I've used the "-n" option on bind 9.1.1 to this
    Tim> affect but can't find the equivilent on bind 8.2.3.

BIND8 is single threaded, so it has no equivalent of BIND9's -n
option. BIND8 can only execute on one processor at a time. Using it on
a multiprocessor will have negligible impact on query throughput. At
best perhaps one CPU runs the kernel and another runs the name
server. If you want a threaded server, you need BIND9. The current
release is 9.1.2. It should be able to handle the above query rate,
though there probably won't be much space capacity.

You might also have to do some stuff to the OS to tune its TCP/IP
stack to cope with that rate of UDP traffic.