[syslog-ng]dns_cache problems with v1.5.13

Balazs Scheidler bazsi@balabit.hu
Thu, 26 Sep 2002 09:42:41 +0200


On Wed, Sep 25, 2002 at 10:58:43AM +0100, colm ennis wrote:
> hi all,
> 
> im currently using syslog-ng version 1.5.13. i have experienced problems
> where negatively cached records were preserved indefinitely. only a full
> restart would cause these ips to be resolved, even a hup would fail.
> 
> i've now turned off caching cos we rely on ips being resolved in these log
> files.
> 
> impt part of the config attached,

Hmm... please check the caching timeouts used. in 1.5.20 the defaults are:

        self->dns_cache_size = 1007;
        self->dns_cache_expire = 3600;
        self->dns_cache_expire_failed = 60;

e.g. failed entries are remembered to 60 seconds. Quickly checking out
test_nscache.c shows that it correctly reresolves entries after the
expire_failed interval elapses:

balabit:~/src/syslog-ng-1.5/src/tests$ ./test_nscache 
initialising nscache: 1000 entries, 3600/1 expiration.
127.0.0.1... adding as 273 (1 iters).
localhost
0.76.56.200... adding as 546 (2 iters).
(null)
0.76.56.200... found on 546: (null) (2 iters, 1033026099/1033026100)
(null)
0.76.56.200... updating entry 546 (1033026104 > 1033026100).
0.76.56.200... adding as 546 (2 iters).
(null)

-- 
Bazsi
PGP info: KeyID 9AF8D0A9 Fingerprint CD27 CFB0 802C 0944 9CFD 804E C82C 8EB1