Failover time to slave nis server? - NFS
This is a discussion on Failover time to slave nis server? - NFS ; Let's say I have 3 linux boxes -- a nis master server, a slave server,
and a client -- where everyone knows about both servers (even the
ypservers map knows about them). Now, I pull the ethernet plug off the
...
-
Failover time to slave nis server?
Let's say I have 3 linux boxes -- a nis master server, a slave server,
and a client -- where everyone knows about both servers (even the
ypservers map knows about them). Now, I pull the ethernet plug off the
master server. How long should I wait until the client starts using the
slave server? What would affect this time if they are all in the same
room and connected to the same switch?
--
Mauricio raub-kudria-com
(if you need to email me, use this address =)
-
Re: Failover time to slave nis server?
On Mon, 16 Oct 2006 09:40:50 -0400, Mauricio Tavares
wrote:
> Let's say I have 3 linux boxes -- a nis master server, a slave server,
>and a client -- where everyone knows about both servers (even the
>ypservers map knows about them). Now, I pull the ethernet plug off the
>master server. How long should I wait until the client starts using the
>slave server? What would affect this time if they are all in the same
>room and connected to the same switch?
Well, clients can cache NIS for a very long time. I know some clients
will cache positive responses for 36,000 seconds while caching
negative responses for 3,600 seconds. So if you're clients have
cached all the auth's they need then it would be 10 hours before they
query the server again.
I assume this is configurable.
I can;t think of anything network-wise that would affect the
time-to-use for the client. As long as the nsswitch.conf file is set
for NIS (do you have local files set?) it should query one, then the
other when it finally wants something.
~F