Using 2 CPUs with BIND

[Mike Diggins]

On Fri, 2 Jan 2009, Stacey Jonathan Marshall wrote:

> Mike Diggins wrote:
>> 
>> Thanks. Would this imply it has detected multiple CPUs? I still don't see 
>> any mention of it in my logs.
> The below output does indicated that you have five threads.  I'm not sure why 
> your log is not showing the message.  Double check the log configuration and 
> check /var/adm/messages for other messages indicating a problem.

Ah ha! It was my logging settings after all. I had to change daemon.notice 
to daemon.info on this line of syslog.conf:

*.err;kern.debug;daemon.info;mail.crit;local3.none;local4.none 
/var/adm/messages

Now I see it has found two CPUs and a bunch of other messages I've never 
seen before. Thanks everyone for the help!

Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.notice] starting 
BIND 9.4.3 -n 2 -c /etc/named.conf
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] found 2 
CPUs, using 2 worker threads
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] using up to 
4096 sockets
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] loading 
configuration from '/etc/named.conf'
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] using 
default UDP/IPv4 port range: [1024, 65535]
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] using 
default UDP/IPv6 port range: [1024, 65535]
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] listening on 
IPv4 interface lo0, 127.0.0.1#53
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] listening on 
IPv4 interface bge0, 130.113.199.8#53
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: 127.IN-ADDR.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: 254.169.IN-ADDR.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: 2.0.192.IN-ADDR.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: 255.255.255.255.IN-ADDR.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: 
0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.IP6.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: 
1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.IP6.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: D.F.IP6.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: 8.E.F.IP6.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: 9.E.F.IP6.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: A.E.F.IP6.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.info] automatic 
empty zone: B.E.F.IP6.ARPA
Jan  2 13:18:49 newblack named[2319]: [ID 873579 daemon.notice] command 
channel listening on 127.0.0.1#953


-Mike



>
> Stace
>> 
>> diggins at newblack<~># /usr/bin/ps -Lp `pgrep named`
>>    PID   LWP TTY        LTIME CMD
>>    605     1 ?           0:00 named
>>    605     2 ?           0:09 named
>>    605     3 ?           0:06 named
>>    605     4 ?           0:34 named
>>    605     5 ?           0:01 named
>> 
>> -Mike
>> 
>> 
>> 
>> On Fri, 2 Jan 2009, Stacey Jonathan Marshall wrote:
>> 
>>> Mike Diggins wrote:
>>>> 
>>>> I noticed that when BIND 9.2.4 on Redhat Linux (Intel x86) starts, the 
>>>> log records:
>>>> 
>>>> dns1 named[28513]: starting BIND 9.2.4 -u named -t /var/named/chroot
>>>> dns1 named[28513]: using 2 CPUs
>>>> 
>>>> When I start BIND on my Solaris 10 SPARC dual CPU (V210) system 9.4.2-P2,
>>>> I don't get the message "using 2 CPUs", but that's what I want. I
>>> The message format changed slightly in BIND 9.4.2-p2, from 
>>> bin/named/main.c:
>>> 
>>> #ifdef ISC_PLATFORM_USETHREADS
>>>     if (ns_g_cpus == 0)
>>>         ns_g_cpus = ns_g_cpus_detected;
>>>     isc_log_write(ns_g_lctx, NS_LOGCATEGORY_GENERAL, NS_LOGMODULE_SERVER,
>>>               ISC_LOG_INFO, "found %u CPU%s, using %u worker thread%s",
>>>               ns_g_cpus_detected, ns_g_cpus_detected == 1 ? "" : "s",
>>>               ns_g_cpus, ns_g_cpus == 1 ? "" : "s");
>>> #else
>>>     ns_g_cpus = 1;
>>> #endif
>>> 
>>> 
>>> 
>>>> compiled it with './configure --prefix=/usr/local/bind --enable-threads'
>>> Take a look at the config.log output to check that threading is indeed 
>>> enabled, it should be by default anyhow on Solaris 10 system.
>>>> and start it with '/usr/local/bind/sbin/named -n 2 -c /etc/named.conf'.
>>>> How do I know it's actually using the two SPARC CPUs?
>>> Use "/usr/bin/ps -Lp `pgrep named`" will show you the number of 
>>> light-weight-processes (LWP), threads, running for the named process. 
>>> Incidently the '-n' option shouldn't be necessary, named can detect the 
>>> number of CPU's on Solaris.
>>> 
>>> Regards,
>>> 
>>> Stace
>>>