Queries for www.tjhinc.com

[Mark Andrews]

> I wrote in part:
> 
> >> One of our users here complained about getting a SERVFAIL when
> >> querying DNS for
> >> 
> >>      www.tjhinc.com.
> >> 
> >> I have done a number of queries (see below), and I do not understand
> >> the results.  Here are the questions:
> >> 
> >> <<Items 1) and 2) omitted here.>>
> >> 
> >> 3) Sometimes I get a response that looks OK; there is one answer
> >>    section and two authority sections with NS records.  Sometimes I
> >>    get a response with one answer section (looks OK) and an
> >>    authority section that has 13 NS records for "com."  Why?
> >>
> >> Here are the dig queries I ran:
> >> 
> >> titania% dig www.tjhinc.com
> >> 
> >> ; <<>> DiG 8.3 <<>> www.tjhinc.com 
> >> ;; res options: init recurs defnam dnsrch
> >> ;; got answer:
> >> ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 2
> >> ;; flags: qr rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 2, ADDITIONAL: 0
> >> ;; QUERY SECTION:
> >> ;;      www.tjhinc.com, type = A, class = IN
> >> 
> >> ;; ANSWER SECTION:
> >> www.tjhinc.com.         1D IN A         64.143.85.46
> >> 
>> ;; AUTHORITY SECTION:
> >> tjhinc.com.             1D IN NS        ns26b.sbc-webhosting.com.
> >> tjhinc.com.             1D IN NS        ns26a.sbc-webhosting.com.
> >> 
> >> ;; Total query time: 54 msec
> >> ;; FROM: titania to SERVER: default -- 146.139.254.5
> >> ;; WHEN: Wed Aug 13 10:29:17 2008
> >> ;; MSG SIZE  sent: 32  rcvd: 103
> >> > titania% !!
> >> dig www.tjhinc.com
> >> 
> >> ; <<>> DiG 8.3 <<>> www.tjhinc.com 
> >> ;; res options: init recurs defnam dnsrch
> >> ;; got answer:
> >> ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 2
> >> ;; flags: qr rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 13, ADDITIONAL: 0
> >> ;; QUERY SECTION:
> >> ;;      www.tjhinc.com, type = A, class = IN
> >> 
> >> ;; ANSWER SECTION:
> >> www.tjhinc.com.         23h59m52s IN A  64.143.85.46
> >> 
> >> ;; AUTHORITY SECTION:
> >> com.                    1d23h59m59s IN NS  f.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  j.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  a.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  d.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  e.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  g.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  k.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  l.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  c.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  m.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  h.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  i.gtld-servers.net.
> >> com.                    1d23h59m59s IN NS  b.gtld-servers.net.
> >> 
> >> ;; Total query time: 2 msec
> >> ;; FROM: titania to SERVER: default -- 146.139.254.5
> >> ;; WHEN: Wed Aug 13 10:29:25 2008
> >> ;; MSG SIZE  sent: 32  rcvd: 272
> >> 
> >> titania# !!
> >> dig www.tjhinc.com
> >> 
> >> ; <<>> DiG 8.3 <<>> www.tjhinc.com 
> >> ;; res options: init recurs defnam dnsrch
> >> ;; got answer:
> >> ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 2
> >> ;; flags: qr rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 2, ADDITIONAL: 0
> >> ;; QUERY SECTION:
> >> ;;      www.tjhinc.com, type = A, class = IN
> >> 
> >> ;; ANSWER SECTION:
> >> www.tjhinc.com.         23h59m54s IN A  64.143.85.46
> >> 
> >> ;; AUTHORITY SECTION:
> >> tjhinc.com.             23h59m54s IN NS  ns26b.sbc-webhosting.com.
> >> tjhinc.com.             23h59m54s IN NS  ns26a.sbc-webhosting.com.
> >> 
> >> ;; Total query time: 1 msec
> >> ;; FROM: titania to SERVER: default -- 146.139.254.5
> >> ;; WHEN: Wed Aug 13 10:30:49 2008
> >> ;; MSG SIZE  sent: 32  rcvd: 103
> >> 
> >> titania#
> 
> And Mark Andrews replied:
> >
> >	BIND 9.5 has a (least recently used) LRU cache with 1009
> >	buckets.  You are seeing the LRU cleaning coming into effect.
> >	There are some bucket related effects visible in the current
> >	code in overmem situtions.
> >
> >	Increase max-cache-size from its default (32M) or reduce
> >	DEFAULT_CACHE_NODE_LOCK_COUNT to reduce the bucket effects.
> >
> >lib/dns/rbtdb.c:
> >#define DEFAULT_CACHE_NODE_LOCK_COUNT   1009    /*%< Should be prime. */
> 
> This is an answer to my questions 1) and 2), but is it an answer to
> question 3)? 

	Yes.  The NS RRset for "sbc-webhosting.com" was not in the
	cache so named added the closest NS RRset it had which was
	"com".

> In the above three queries, I see
>
>      a) a good query with a TTL of 1D and tjhinc.com authority,
>      b) then a query with a slightly shorter than 1D TTL but with .com
>              authority,
>      c) then a good query with a slightly shorter than 1D TTL and
>              tjhinc.com authority.
> 
> Another question arises - if the default max-cache-size is 32M
> (and titania has 2G of memory), and the cache is supposedly full,
> then why is the named_dumpdb file I created yesterday only 818340
> bytes long?

	named_dumpdb will usually be smaller than the memory used
	internally.

	If you are going to stay w/ 32M I would used one of these
	values for DEFAULT_CACHE_NODE_LOCK_COUNT (17 19 23 29 31).

	Mark

> ----------------------------------------------------------------------
> Barry S. Finkel
> Computing and Information Systems Division
> Argonne National Laboratory          Phone:    +1 (630) 252-7277
> 9700 South Cass Avenue               Facsimile:+1 (630) 252-4601
> Building 222, Room D209              Internet: BSFinkel at anl.gov
> Argonne, IL   60439-4828             IBMMAIL:  I1004994
> 
-- 
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: Mark_Andrews at isc.org