On Mon, Feb 19, 2007 at 10:23:29AM -0500, lists wrote:
> I have a problem where tinydns & dnscache no longer respond when I have
> more than a certain number of domains in my data file. The offending
Can you be more exact in your description of the problem. tinydns
doesn't respond, or dnscache doesn't respond? Given what setup and what
queries? Can you show (in detail) the steps you performed, what you
expected, and what you saw? Can you show us the contents of your
/service/dnscache and /service/tinydns directories?
> data.cdb has a size of 1.37 MB. It's definitely a size issue, since it
> [...]
> I've reset the CACHESIZE to 100000000 and the DATALIMIT to 104857600 as
> described at http://cr.yp.to/djbdns/cachesize.html . I've also set the
data.cdb is completely unrelated to dnscache; changing your CACHESIZE
and DATALIMIT for dnscache won't change anything.
> softlimit to -d1000000 in /service/tinydns/run . I previously had it all
> set to defaults.
The data.cdb file is mmap()ed, which shouldn't count against the
datalimit. This shouldn't matter (though I suppose it's possible that
you have some weird kernel patch which operates in this way; you can try
setting the softlimit to something very high).
> The log The log /service/dnscache/log/main/current reports no errors when
> I'm having this problem, though the first time I had this problem back in
> January, it said servfail mail.etemad.org. input/output error .
What _exactly_ does the log report (from dnscache, and from tinydns)?
Are the requests being received? Are any log entries created for them?
-Peff
|