rndc status field meaning please
cathya at isc.org
Tue Jul 21 13:10:17 UTC 2015
I don't think we do document the output from "rndc status" explicitly
line by line in the BIND Administrator Manual, so I'll respond to your
questions below, and I'll see about getting the documentation updated.
For anything else you need to know, please refer to the manuals
On 21/07/2015 07:57, 이윤호 wrote:
> CPUs found: 4
This is the number of processors that BIND detected. This may be more
that the number of physical CPUs due to hyperthreading and similar.
> worker threads: 4
When named starts up as a multi-threaded build, it has 3 threads with
specific roles, plus 'n' worker threads. The worker threads handle
'tasks'. A task is 'something that BIND has to do', for example, start
processing a new inbound request, send a query to another server,
process a reply from another server, and so on.
If it helps, think of a running named process as being like an OS all by
itself. The worker threads are like CPUs. Tasks ('things to be done')
are equivalent to processes. Tasks get queued when they are ready to be
launched, and then will get executed by one of the worker threads.
By default, n is set to the number of CPUs that named detected, but you
can also control this when launching named by using the -n option.
The -n option is documented in the ARM:
> UDP listeners per interface: 2
A UDP listener is another example of a 'task'.
By default, we start named with n/2 listeners per interface. There is a
bigger discussion of this here:
> number of zones: 16
This is the number of zones that named is authoritative for. This
includes automatic and internal-only zones, so it is going to be a
bigger number than the count of zone statements in your named.conf.
Look at the log of named starting up to see what you have.
> xfers running: 0
This is the number of zone transfers currently running.
> xfers deferred: 0
This is the number of zone transfers queued (waiting to run).
> soa queries in progress: 0
When named initiates a 'refresh' of a slave zone, either because the
refresh timer has popped, or it has received a notify from a master,
then it sends SOA queries to the masters for the zone to check their
serial number. This shows how many of those queries are currently
> recursive clients: 0/9900/10000
The first number is the count of client queries that are waiting to be
answered while named is sending queries to other servers to get the
answers (in other words, it it resolving the client queries recursively).
The second number is the soft limit. When named reaches this number of
recursive clients, then it will still accept the new one, but at the
same time it will drop the oldest one that is still waiting.
The third number is the hard limit. If named reaches this number then
it cannot accept a new client query that requires recursive resolution.
> tcp clients: 0/100
Client connections using TCP.
> bind!! you have rndc manual?
We have man pages and we have the Administrator Reference Manual.
More information about the bind-users