Results 1 to 3 of 3
Hi, I have set up a caching name server with bind/named on my fedora box. I have a root.hints too. When i run dig "some url" I get a response ...
Enjoy an ad free experience by logging in. Not a member yet? Register.
- 04-17-2012 #1
- Join Date
- May 2006
caching name server cache expiration
I have set up a caching name server with bind/named on my fedora box. I have a root.hints too. When i run dig "some url" I get a response in a couple of 100's msec. When dig the same url again it comes back in zero or 1-2 msec. All good so far, the server is remembering. The problem I have is that it seems to then forget about 10-15 minutes later, we are back to the 100's msec in response to dig requests to the same url's. This also occurs with my LAN hostnames too. Slow response, fast response on the second request, leave it for 10-15 minutes and slow again.
Is there a way to increase the time the cache is kept for, say a month or so?
- 04-17-2012 #2
Imho: Dont overwrite the TTL of cached records, even if you should find a -in that way- broken dnserver and/or a patch for bind.
The zone admin has set the time-to-live for a reason.
For example, there might be a round-robin approach to loadbalance requests to multiple servers.
Or a ttl has been reduced to accomodate a migration.
A modified ttl in such situations will almost certainly lead to false conclusions in case of an issue.
For your zones in your lan, you can of course set the ttl to a value you like.
Last edited by Irithori; 04-18-2012 at 01:37 PM.You must always face the curtain with a bow.
- 04-18-2012 #3
I agree you should not mess with the TTL's as the admin usually sets them to the time he wants for a reason. That being said I have seen a lot of stupid DNS admins who don't have a clue and set the TTL's to something very short times.
A normal setting would be a day. System admins that are worth their weight will know how to setup a system that allows for a longer TTL and allow for outages.
The adventure of a life time.
Linux User #296285