Re: difference in dnscache installs

On Tue, 19 Jan 2010 12:15:10 +0000
James Sutherland <[email protected]> wrote:

> On Tue, Jan 19, 2010 at 9:54 AM, richard lucassen
> <[email protected]> wrote:
> >
> > The thing is that I have around 180 dnscache running and none of
> > these show this problem. Only these two. The only difference is:
> >
> > cat DATALIMIT CACHESIZE
> > 44000000
> > 40000000
> >
> > These two instances of dnscache are quite busy indeed.
> 
> That sounds like a possible culprit: if these names take 12 steps of
> recursion to resolve, presumably it's taking quite a long *time* to
> resolve them - so if the cache is busy, it may be giving up on that
> query while it's still part way through chasing all the levels of
> recursion. Are you seeing any "drop" entries in the cache logs on the
> affected machines?

I think you're right, I looked up "www.singaporeair.com":

# tail -F main/current | tai64nlocal | grep www.singaporeair.com

2010-01-19 14:32:01.567387500 query 273 ac100006:9b96:29b0 1
www.singaporeair.com.

2010-01-19 14:32:01.567535500 tx 0 1 www.singaporeair.com. . ca0c1b21
c1000e81 803f0235 80080a5a c0702404 c0cbe60a c6290004 c021040c c707532a
c0e44fc9 c03a801e c0249411 c00505f1

2010-01-19 14:32:01.587689500 tx 0 1 www.singaporeair.com. com.
c00c5e1e c029a21e c005061e c037531e c0210e1e c036701e c01f501e c0304f1e
c01a5c1e c034b21e c023331e c02bac1e c02a5d1e

2010-01-19 14:32:01.608783500 tx 0 1 www.singaporeair.com.
singaporeair.com. cb8f8804 cb8f8805

2010-01-19 14:32:06.571815500 query 380 ac100006:9bcb:29b0 1
www.singaporeair.com.

2010-01-19 14:32:06.571936500 tx 0 1 www.singaporeair.com. . 80080a5a
c1000e81 c0cbe60a c0702404 c03a801e c021040c 803f0235 c00505f1 c0249411
c0e44fc9 c6290004 c707532a ca0c1b21

2010-01-19 14:32:06.673224500 tx 0 1 www.singaporeair.com. com.
c037531e c02bac1e c01a5c1e c01f501e c0304f1e c036701e c029a21e c023331e
c00c5e1e c02a5d1e c0210e1e c034b21e c005061e

2010-01-19 14:32:07.021175500 tx 0 1 www.singaporeair.com.
singaporeair.com. cb8f8805 cb8f8804

2010-01-19 14:32:11.888564500 servfail www.singaporeair.com.
input/output error

If dnscache doesn't get it's response within 10 seconds apparently it
reports servfail. It's a 1Gb internet line, which is not overloaded.

Would it be wise to increase this timeout somewhere in the source? AFAIK
there is no environment variable. I have set the cachesize to 100MB
now, but no improvement (just to be sure)

The main question remains why these two instances of dnscache show this
phenomenon while others on much smaller DSL lines do not.

R.

-- 
___________________________________________________________________
It is better to remain silent and be thought a fool, than to speak
aloud and remove all doubt.

+------------------------------------------------------------------+
| Richard Lucassen, Utrecht                                        |
| Public key and email address:                                    |
| http://www.lucassen.org/mail-pubkey.html                         |
+------------------------------------------------------------------+