1) Is it possible that the same server periodically changes its key fingerprint?
Yes, the server key(s) may in theory be changed at any time by the administrator of that server. However, that destroys established trust, so it should not be done in practice without reason.
It can occur if a server is rebuilt (new hardware without preserving ssh server keys, or nuke from orbit scenario). It can also occur if there are multiple servers with different keys which use the same hostname (behind a load balancer, for example).
In addition, ssh servers typically have multiple host keys of different types (rsa, dsa, ecdsa, ed25519...) so you may see a different key fingerprint when using one client as compared to another client, depending on what key exchange is negotiated. However, any given client should always get the same key from the server if neither client nor server configuration changes.
2) Is there a way to suppress the fingerprint check only for local addresses, that is in the range 192.168.1.1-254
Yes. From the ssh_config manual page, the options
StrictHostKeyChecking affect this behaviour. To suppress these checks for all hostnames beginning "192.168.1." followed by one to three characters, use
Note that the patterns supported in the
Host clause are not full regular expressions or shell globs, but string matches with the single character
? and the greedy
* wildcards only. Hence, if someone managed to register
1.com, then this pattern would also match the remote server
Using the greedy glob would make a simpler configuration clause
Host 192.168.1.*, but also allow matches to a wider range of mixed alpha-numeric names, such as
192.168.1.evil.qwerty.org. The chances of inadvertently connecting (without host key verification) to such a server is very low; however it is good to be aware that this is simply a string glob and not something that will necessarily be parsed as a subnet address, just because it looks somewhat like one.