0

We are running Bind (9.8) as Caching Nameserver on a CentOS 6.x based mail server. We get frequent domain not found error for certain domains resulting in mail delivery failure. It was discovered that all those domains have low TTL values (300-1800) for the MX records and the nameserver gives a blank answer for the MX look up query after the TTL expiry. If the name server is restarted, a proper reply is received. Ideally, the Caching Nameserver should do a fresh look up in case the cached records expire but it is not doing so in case of these domains. What can be done to fix this problem since its practically not possible to restart the caching name server every five minutes.

Thanks,

catman
  • 41
  • 1
  • 6
  • I think more info will be required. Are the authoritative servers actually providing consistent answers for one thing? Can you give an example domain for which you experience this problem? – Håkan Lindqvist Feb 13 '15 at 07:39
  • Some of these domains are: sbi.co.in, gabriel.co.in, hdfcbank.com – catman Feb 13 '15 at 09:38
  • And yes, the answers seem to be consistent. – catman Feb 13 '15 at 12:21
  • You could run "rndc dumpdb -cache" to get a picture of the name server's cached lookups. Compare the NS records for the relevant domains when resolution works with those when it fails. – Andy Feb 14 '15 at 09:24
  • There was no difference. – catman Feb 15 '15 at 06:29

0 Answers0