I have a couple of test systems running Centos 6 and I thought I'd try out Centos 7. Apart from all the obvious differences (no iptables, no ifconfig), there is one thing that really had me scratching my head. On centos 6, my hostname looks like this:
# hostname centos6
On centos 7, my hostname needs to look like this:
# hostname centos7.fqdn.foo
If I set the hostname to simply
centos7, this causes DNS issues and I am only able to resolve external addresses, such as google.com, but no internal hostnames from our internal DNS server. Since there does not seem to be a requirement on my centos6 box to have this "clunky" hostname, I am wondering if something changed between releases, or if I'm simply overlooking some configuration made on centos 6 (such as a DOMAIN variable in the network configuration or something like that) that is missing on the centos 7 box.
What also confuses me is that on the centos 7 box, this command works:
# ping centos6 PING centos6.fqdn.foo (192.168.1.15) 56(84) bytes of data. 64 bytes from centos6.fqdn.foo (192.168.1.15): icmp_seq=1 ttl=64 time=0.399 ms
So apparently the ping command can translate the short host name into the long one, but when I try the same with nslookup:
# nslookup centos6 ;; Got SERVFAIL reply from 192.168.1.11, trying next server
The short host name is obviously not enough. Using the FQDN with nslookup works fine.
I am assuming that the problem here is that as soon as my centos 7 box leaves the
fqdn.foo domain, DNS breaks down. I just don't understand why, and why this doesn't seem to be the case on centos 6.
edit This may have been as easy as running
nmtui and making sure both DNS and search were filled out for the proper ethernet adapter. I was almost sure I had done it before but perhaps I forgot to do
service network restart, who knows. It works now anyway.