▸ quoted from Henrik Størner
On Thu, 2006-06-08 at 22:28 +0200, Henrik Stoerner wrote:
My hope is that this will be a rather easy thing to do, once I
get around to rewriting the network tester so that it can engage
in a real dialogue. Currently the network tester connects,
sends something, and grabs the response. Period. It won't look
at the response and go "hey, I need to send another request"
and then do that. This is also causing problems for e.g. SMTP
checks, because the current default Hobbit setup is actually
not SMTP protocol compliant - and a few SMTP servers out there
actually complain about it, causing Hobbit to think they are
out of order, when in fact they are running fine.
So - the plan is to put some more advanced communications
mechanism into the network tester, and with that in place it
should be easier to provide the redirect-support you need.
I have been thinking recently (well, since I screwed up the pop server
last weekend such that it passed the hobbit pop test, but no user could
collect any email), that it would be really nice to extend the network
tests to engage in a "real" test.
ie, the SMTP test would actually send an email
the POP test would actually retrieve an email
etc...
Once, many moons ago, I did write some utils to do these things with BB,
in fact, it actually went one step further. The email which was sent via
SMTP had to show up on the POP test within a certain time frame or else
it would also alarm. Since we had a seperate farm for POP and SMTP, this
meant we could confirm there were no delays between the two sides etc...
Just my 5c worth...
If I get some time, I'll go and write some ext scripts to do this stuff,
but it doesn't look likely at the moment.
However, I have recently added tinydns statistics graphing recently, so
here is how it was done:
Install tinystats as per the installation instructions
I'm still using BB client on my two old DNS servers and hobbit on the
two new ones, so this ext script works on both:
tinystats.sh
#!/bin/bash
CAT=/bin/cat
CUT=/usr/bin/cut
STATFILE=/var/dns/tinydns/log/main/tinystats.out
TMPSTAT=$BBTMP/tinystats.$$
TMPSTATout=$BBTMP/tinystats.out.$$
TEST=dnsnum
${TAIL} -1 ${STATFILE} > ${TMPSTAT}
${HEAD} -1 ${STATFILE} >> ${TMPSTAT}
for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
do
stats="`${CAT} ${TMPSTAT}`"
output=`echo "$stats" | ${CUT} -d: -f${i}`
echo ${output} | ${TR} " " ":" >> $TMPSTATout
done
${BB} ${BBDISP} "status ${MACHINE}.${TEST} green `${DATE}` DNS
Statistics
`cat ${TMPSTATout}`"
${RM} -f ${TMPSTAT} ${TMPSTATout}
PS, it is a crap script, but it works for me, if anyone can suggest a
better way to pull out the data, I'd appreciate it :)
add "dnsnum=ncv" to the end of your TEST2RRD string in hobbitserver.cfg
add "dnsnum" to the end of your GRAPHS string in hobbitserver.cfg
add the following to your hobbitgraph.cfg
[dnsnum]
TITLE TinyDNS Utilization
YAXIS Stats
DEF:A=dnsnum.rrd:a:AVERAGE
DEF:NS=dnsnum.rrd:ns:AVERAGE
DEF:CNAME=dnsnum.rrd:cname:AVERAGE
DEF:SOA=dnsnum.rrd:soa:AVERAGE
DEF:PTR=dnsnum.rrd:ptr:AVERAGE
DEF:MX=dnsnum.rrd:mx:AVERAGE
DEF:TXT=dnsnum.rrd:txt:AVERAGE
DEF:AAAA=dnsnum.rrd:aaaa:AVERAGE
DEF:ANY=dnsnum.rrd:any:AVERAGE
DEF:OTHER=dnsnum.rrd:other:AVERAGE
DEF:NOTAUTH=dnsnum.rrd:notauth:AVERAGE
DEF:NOTIMPL=dnsnum.rrd:notimpl:AVERAGE
DEF:BADCLASS=dnsnum.rrd:badclass:AVERAGE
LINE2:A#0000FF:A
LINE2:NS#000088:NS
LINE2:CNAME#00FF00:CNAME
LINE2:SOA#008800:SOA
LINE2:PTR#FF0000:PTR
LINE2:MX#880000:MX
LINE2:TXT#FF00FF:TXT
LINE2:AAAA#880088:AAAA
LINE2:ANY#FFFF00:ANY
LINE2:OTHER#555555:O
LINE2:NOTAUTH#0088FF:NOTAUTH
LINE2:NOTIMPL#008888:NOTIMPL
LINE2:BADCLASS#00FFFF:BADCLASS
COMMENT:\n
GPRINT:A:LAST:A \: %5.1lf (cur)
GPRINT:A:MAX:A \: %5.1lf (max)
GPRINT:A:MIN:A \: %5.1lf (min)
GPRINT:A:AVERAGE:A \: %5.1lf (avg)\n
GPRINT:ANY:LAST:ANY \: %5.1lf (cur)
GPRINT:ANY:MAX:ANY \: %5.1lf (max)
GPRINT:ANY:MIN:ANY \: %5.1lf (min)
GPRINT:ANY:AVERAGE:ANY \: %5.1lf (avg)\n
GPRINT:MX:LAST:MX \: %5.1lf (cur)
GPRINT:MX:MAX:MX \: %5.1lf (max)
GPRINT:MX:MIN:MX \: %5.1lf (min)
GPRINT:MX:AVERAGE:MX \: %5.1lf (avg)\n
GPRINT:OTHER:LAST:OTHER \: %5.1lf (cur)
GPRINT:OTHER:MAX:OTHER \: %5.1lf (max)
GPRINT:OTHER:MIN:OTHER \: %5.1lf (min)
GPRINT:OTHER:AVERAGE:OTHER \: %5.1lf (avg)\n
GPRINT:PTR:LAST:PTR \: %5.1lf (cur)
GPRINT:PTR:MAX:PTR \: %5.1lf (max)
GPRINT:PTR:MIN:PTR \: %5.1lf (min)
GPRINT:PTR:AVERAGE:PTR \: %5.1lf (avg)\n
GPRINT:SOA:LAST:SOA \: %5.1lf (cur)
GPRINT:SOA:MAX:SOA \: %5.1lf (max)
GPRINT:SOA:MIN:SOA \: %5.1lf (min)
GPRINT:SOA:AVERAGE:SOA \: %5.1lf (avg)\n
BTW, that needs further adjustment, I'd like to split it into two
graphs, one for error, and one for 'normal'. Also, it doesn't show data
for the various stats that all showed as 0 on my servers. Also, you may
want more GPRINT lines, I just copied/modified the bind ones....
That it all....
Apart from the above graphing things, the only other thing I haven't
worked out/looked up is how to get the graph to show up on the dnsnum
column, currently it is only on the trends column.
Hope that helps someone else out there.
Regards,
Adam