Prev: Crippled newfs in Solaris 10?
Next: Adding Desktop Configuration Manager to Java Web Console
From: weathercoach on 17 Jan 2007 12:29

Hello.
I am installing Solaris 10 (11/06, Sparc) on a v100 via jumpstart and
cannot get the installation to be completely hands off as I get the
much discussed nfs4 domain prompt

This system is configured with NFS version 4, which uses a
domain name that is automatically derived from the system's
name services. The derived domain name is sufficient for most
configurations. In a few cases, mounts that cross different
domains might cause files to be owned by nobody due to the
lack of a common domain name.

Do you need to override the system's default NFS verion 4 domain
name (yes/no)? [no]


My understanding is that in solaris 10 11/06 the use of the sysidcfg
keyword "nfs4_domain" will
alleviate the prompts but that has not worked for me. (Note: I have
tried using the value of "dynamic" and the value of "sub.domain.net"
without any perceptable differences).

#cat sysidcfg

system_locale=en_US
timezone=US/Pacific
terminal=vt100
timeserver=localhost
name_service=DNS {domain_name=sub.domain.net
name_server=10.17.8.25}
root_password=1ixRgD.QH7mIQ
network_interface=dmfe0 {hostname=lig42
default_route=10.17.8.1
ip_address=10.17.8.40
netmask=255.255.255.0
protocol_ipv6=no}
security_policy=NONE
nfs4_domain=dynamic

After the client jumpstart some related files look like this

# more /etc/resolv.conf
domain sub.domain.net
nameserver 10.17.8.25

# tail -1 /etc/default/nfs
#NFSMAPID_DOMAIN=domain

# grep -i hosts: /etc/nsswitch.conf
hosts: files dns

# cat /var/run/nfs* ( i dont know if this file get's populated
before or after the nfs4 prompt)
sub.domain.net

I have setup the DNS txt record _nfsv4idmapdomain but during the
jumpstart process
the client never makes a query for this record? The usual client
lookups for itself do come across
the wire to the name server but nothing for the nfsv4 txt record?

dig +domain=sub.domain.net -t TXT _nfsv4idmapdomain

; <<>> DiG 9.2.4 <<>> +domain=sub.domain.net -t TXT _nfsv4idmapdomain
;; global options: printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 224
;; flags: qr rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 0, ADDITIONAL: 0

;; QUESTION SECTION:
;_nfsv4idmapdomain.sub.domain.net. IN TXT

;; ANSWER SECTION:
_nfsv4idmapdomain.sub.domain.net. 30 IN TXT "sub.domain.net"

;; Query time: 5 msec
;; SERVER: 172.17.8.25#53(10.17.8.25)
;; WHEN: Wed Jan 17 08:25:28 2007
;; MSG SIZE rcvd: 75

# ls -al /etc/.NFS*
-rw-r--r-- 1 root root 0 Jan 17 08:23
/etc/.NFS4inst_state.domain

# more /etc/.sysID*
1 # System previously configured?
1 # Bootparams succeeded?
1 # System is on a network?
1 # Extended network information gathered?
1 # Autobinder succeeded?
1 # Network has subnets?
1 # root password prompted for?
1 # locale and term prompted for?
1 # security policy in place
vt100


My understanding is that using the nfs4_domain keyword in the
sysidcfg file would have
added the following line to the /etc/.sysIDtool.state file

1 # NFSv4 domain configured


Anyhoo I do realize that this prompt can be overcome through the use
of finish scripts but I would like to understand why the installation
is dropping to an interactive prompt since I *think* I followed al of
the instructions to disable the prompt. Obviously I've missed (or
misunderstood) something and would appreciate any inputs.

TIA. Walter.

 | 
Pages: 1
Prev: Crippled newfs in Solaris 10?
Next: Adding Desktop Configuration Manager to Java Web Console