Re: Missing rpc in new install. - Solaris

This is a discussion on Re: Missing rpc in new install. - Solaris ; On Wed, 15 Feb 2006 20:46:30 -0700, GreyCloud wrote: > I notice that in rc2.d, the S71rpc startup script is missing. Why would you expect that script to exist? Those functions have been moved to SMF in Solaris 10....

+ Reply to Thread
Results 1 to 10 of 10

Thread: Re: Missing rpc in new install.

  1. Re: Missing rpc in new install.

    On Wed, 15 Feb 2006 20:46:30 -0700, GreyCloud wrote:

    > I notice that in rc2.d, the S71rpc startup script is missing.


    Why would you expect that script to exist? Those functions have been
    moved to SMF in Solaris 10.


  2. Re: Missing rpc in new install.

    On Thu, 16 Feb 2006 12:24:41 -0700, GreyCloud wrote:

    > Dave Uhring wrote:
    >> On Wed, 15 Feb 2006 20:46:30 -0700, GreyCloud wrote:
    >>
    >>
    >>>I notice that in rc2.d, the S71rpc startup script is missing.

    >>
    >>
    >> Why would you expect that script to exist? Those functions have been
    >> moved to SMF in Solaris 10.
    >>

    >
    > I was googling for a particular problem to DT startup error message that
    > when logging in as root under CDE. It won't let you do anything else
    > but throws you back to the login screen. So I logged into failsafe.
    > The article suggested that S71rpc is missing. Now how that happens I
    > don't know. The main thing tho, is that there isn't any internet
    > access. So the actual context of the installer screens comes into question.


    Solaris 9 and earlier versions have that script; Solaris 10 does not. To
    see if the rpc services have been started:

    [/]# svcs -a | grep rpc
    disabled Jan_10 svc:/network/rpc/keyserv:default
    disabled Jan_10 svc:/network/rpc/nisplus:default
    disabled Jan_10 svc:/network/rpc/bootparams:default
    disabled Jan_10 svc:/network/rpc/rex:default
    disabled Jan_10 svc:/network/rpc/spray:default
    disabled Jan_10 svc:/network/rpc/wall:default
    disabled Jan_10 svc:/network/rpc/ocfserv:default
    online Jan_10 svc:/network/rpc/bind:default
    online Jan_10 svc:/network/rpc/gss:default
    online Jan_10 svc:/network/rpc/mdcomm:default
    online Jan_10 svc:/network/rpc/meta:default
    online Jan_10 svc:/network/rpc/metamed:default
    online Jan_10 svc:/network/rpc/metamh:default
    online Jan_10 svc:/network/rpc/rstat:default
    online Jan_10 svc:/network/rpc/rusers:default
    online Jan_10 svc:/network/rpc/smserver:default
    online Jan_10 svc:/network/rpc-100235_1/rpc_ticotsord:default
    online Jan_10 svc:/network/rpc-100083_1/rpc_tcp:default
    online Jan_10 svc:/network/rpc-100068_2-5/rpc_udp:default


    > I may have to do a sys-unconfig and start over again. I'm behind a
    > cisco router that does DHCP, so I don't need that in my solaris box. A
    > fixed address is fine for a LAN. It seems that the installer does not
    > setup the files appropriately for internet access, which I figure it
    > should do. Nor does it ask for my ISPs name or IP address.


    If you were not asked for your IP address it is likely that you are using
    a NIC which is unsupported in the base system. Does the output of
    `ifconfig -a` show that NIC? An example:

    [/]# ifconfig -a
    lo0: flags=2001000849 mtu 8232 index 1
    inet 127.0.0.1 netmask ff000000
    iprb0: flags=1000843 mtu 1500 index 2
    inet 192.168.0.5 netmask ffffff00 broadcast 192.168.0.255
    ether 0:8:c7:45:41:ba

    lo0 is the loopback interface; iprb0 is an Intel 82558 NIC.


  3. Re: Missing rpc in new install.

    On Fri, 17 Feb 2006 11:18:41 -0700, GreyCloud wrote:

    > Dave Uhring wrote:


    >> If you were not asked for your IP address it is likely that you are using
    >> a NIC which is unsupported in the base system.

    >
    > It showed a machine picked IP address to use.


    It's been a while since I installed Solaris but IIRC you were asked
    whether you wished to use DHCP or assign a static address. It appears
    that you *chose* to use DHCP.

    >> Does the output of
    >> `ifconfig -a` show that NIC? An example:
    >>

    >
    > It is a supported NIC... Intel Pro/100.
    > It worked fine under Solaris 8.


    Then it uses the iprb driver and should be identified as iprb0.

    > The problem occurs when I login as root to CDE.


    Then do not do that! You have *no* requirement ever to log in as root.
    You can create an unprivileged user account during the installation
    process:

    # useradd -c 'GreyCloud' -d /export/home/mist -m -s /usr/bin/bash mist
    # passwd mist

    > Once inside is where
    > the problems are. It won't let you do anything except click on the
    > dialog box 'OK' and back to the login screen.
    > Loggin into the Java Desktop works ok, but no internet access.


    We need to see the outputs of `ifconfig -a`, `netstat -rn`, `cat
    /etc/resolv.conf` and `grep hosts /etc/nsswitch.conf`.


  4. Re: Missing rpc in new install.

    On Fri, 17 Feb 2006 12:58:18 -0700, GreyCloud wrote:

    > Dave Uhring wrote:


    >> It's been a while since I installed Solaris but IIRC you were asked
    >> whether you wished to use DHCP or assign a static address. It appears
    >> that you *chose* to use DHCP.

    >
    > It doesn't ask that anymore. After choosing 'None' under the networking
    > options of NFS, NIS, LDAP, DNS, and NONE, I was assigned a static IP
    > address. Another option was to discover the default router on reboot.


    You are not referring to "networking options" but name service options.
    This question has nothing to do with the assignment of IP address.

    As for the default router, your router may or may not provide the required
    information. Easier and more foolproof to define the default route
    yourself.

    > The install program no longer asks for a regular user account.


    It never has asked that, at least since Solaris 2.5.1.

    > That's why I have to login initially as root to set one up in the first
    > place. I don't know if admintools is still around in Solaris 10.


    Admintool is gone. Either create the user account during installation or
    log in using the Failsafe option and create the account just as I
    indicated below:

    >> # useradd -c 'GreyCloud' -d /export/home/mist -m -s /usr/bin/bash mist
    >> # passwd mist


    > netstat -rn shows:
    >
    > Routing Table: IPv4
    > Destination Gateway Flags Ref Use Interface
    > ----------------------------------------------------------------------
    > 192.168.1.0 192.168.1.104 U 1 0 iprb0
    > 224.0.0.0 192.168.1.104 U 1 0 iprb0
    > default 192.168.1.1 UG 1 0
    > 127.0.0.1 127.0.0.1 UH 3 154 lo0


    Apparently you are using an inexpensive SOHO router which provides DHCP
    service. Unless you are using a notebook machine which is to be used on
    several different networks it is best to configure a static IP address in
    the range of 192.168.1.2 - 192.168.1.99.

    At least you do have a default route. You should be able to ping external
    hosts:

    $ ping 216.109.112.135

    > cat /etc/resov.conf:
    >
    > domain bresnan.net
    > nameserver 69.145.251.246


    The nameserver entries should be:

    nameserver 69.145.248.3
    nameserver 69.145.232.3

    > grep hosts /etc/nsswitch.conf:
    >
    > hosts: files


    Which means that gethostbyaddress() *never* uses dns.

    # cp /etc/nsswitch.dns /etc/nsswitch.conf

    The hosts: line must be

    hosts: files dns

    I suggest that you execute sys-unconfig and reconfigure using a static
    address as above and select NONE as you name service. After the system
    comes back up to run level 3 then configure /etc/resolv.conf and
    /etc/nsswitch.conf as I indicated.


  5. Re: Missing rpc in new install.

    On Fri, 17 Feb 2006 17:18:12 -0700, GreyCloud wrote:

    > Ok, I finally got this up and running on the net.
    > I had to do a print out from OS X on its /etc/hosts file.
    >
    > domain hosts.bc1.bresnan.net
    > nameserver 69.146.17.3
    > nameserver 69.145.248.4
    > nameserver 69.146.17.2


    OS-X is based on FreeBSD-3.2. That file is most likely /etc/resolv.conf,
    not /etc/hosts.

    > I don't know what UNIX command you need to issue to get the actual
    > domain name, nor getting the nameserver addresses. Somehow, OS X does it
    > automatically during setup.


    I don't know what OS-X command one would use. With Solaris you are
    expected to know certain things before configuring your networking,
    information which you could have obtained from the OS-X host.

    > But it did get me on the net.


    So I see.

    > Now I need to get rid of the annoying 'host GreyCloud unknown; sleeping
    > for retry' message.


    Since your domain appears to be hosts.bc1.bresnan.net, edit
    /etc/inet/hosts and /etc/inet/ipnodes to provide the fully qualified
    hostname at the line where your IP address is located.

    192.168.0.7 tarfu.localnet.tld tarfu loghost

    The message is issued by sendmail, which is unable to resolve your FQHN.


  6. Re: Missing rpc in new install.

    On Fri, 17 Feb 2006 22:21:49 -0700, GreyCloud wrote:

    > Dave Uhring wrote:


    >> Since your domain appears to be hosts.bc1.bresnan.net, edit
    >> /etc/inet/hosts and /etc/inet/ipnodes to provide the fully qualified
    >> hostname at the line where your IP address is located.
    >>
    >> 192.168.0.7 tarfu.localnet.tld tarfu loghost
    >>
    >> The message is issued by sendmail, which is unable to resolve your FQHN.
    >>

    >
    > It seems that neither Sol 8 or 10 ever gets this right during the
    > install phase.


    Solaris expects that you have a nameserver on your network which will
    resolve your hostname and IP address both forward and inversely. If you
    choose not to run a local DNS server then you need to fix "files".


  7. Re: Missing rpc in new install.

    On Sat, 18 Feb 2006 10:51:22 -0700, GreyCloud wrote:

    > Dave Uhring wrote:


    >> Solaris expects that you have a nameserver on your network which will
    >> resolve your hostname and IP address both forward and inversely. If you
    >> choose not to run a local DNS server then you need to fix "files".


    > I wished they'd mention this right off. I wonder how many free
    > downloads that ended up on solitary workstations vs. networked ones?


    No idea but I expect that most of those downloads were by Windows users
    who know *nothing* about UNIX and their CD-Rs were long ago sent to file
    13. Since I run DNS servers on my networks I have no such problems.

    > Any way, another problem is still there, but nothing to do with getting
    > on the internet. It has to do with the session choices. You can't use
    > CDE without fixing some files that are missing. ???


    You would have been notified at the end of the installation had some
    packages failed to install. Otherwise there is no good reason that CDE
    would fail to start.

    Since you have a new installation remove the ~/.dt directories and try
    again. Otherwise post the list of missing files or the actual error
    messages.


  8. Re: Missing rpc in new install.

    On Sat, 18 Feb 2006 21:20:56 -0700, GreyCloud wrote:

    > Dave Uhring wrote:


    >> You would have been notified at the end of the installation had some
    >> packages failed to install. Otherwise there is no good reason that CDE
    >> would fail to start.

    >
    > CDE starts, it also shows the first time startup window and the files
    > window. But there is also a message window that warns that three files
    > need to be checked. The only action is to click on the OK button, which
    > leads back to the login screen.


    If we knew *which* files we could advise which packages need to be
    reinstalled. Otherwise it appears that you get to use the Microsfot
    solution - completely reinstall the OS taking careful note of any error
    messages.

    > Another error inside that file refers to "ld.so.1: dtappgather: fatal:
    > /usr/lib/libC.so.5: hardware capability unsupported: 0x1 [FPU]"
    >
    > Huh? both these libs are 32-bit ELF binaries.


    The FPU is the floating point processor. What CPU are you using?


  9. Re: Missing rpc in new install.

    GreyCloud wrote:

    > Dave Uhring wrote:
    >
    >> On Fri, 17 Feb 2006 22:21:49 -0700, GreyCloud wrote:
    >>
    >>
    >>> Dave Uhring wrote:

    >>
    >>
    >>
    >>>> Since your domain appears to be hosts.bc1.bresnan.net, edit
    >>>> /etc/inet/hosts and /etc/inet/ipnodes to provide the fully qualified
    >>>> hostname at the line where your IP address is located.
    >>>>
    >>>> 192.168.0.7 tarfu.localnet.tld tarfu loghost
    >>>>
    >>>> The message is issued by sendmail, which is unable to resolve your
    >>>> FQHN.
    >>>>
    >>>
    >>> It seems that neither Sol 8 or 10 ever gets this right during the
    >>> install phase.

    >>
    >>
    >>
    >> Solaris expects that you have a nameserver on your network which will
    >> resolve your hostname and IP address both forward and inversely. If you
    >> choose not to run a local DNS server then you need to fix "files".
    >>

    >
    > I wished they'd mention this right off. I wonder how many free
    > downloads that ended up on solitary workstations vs. networked ones?
    >
    > Any way, another problem is still there, but nothing to do with getting
    > on the internet. It has to do with the session choices. You can't use
    > CDE without fixing some files that are missing. ???
    >
    >

    It was obvious to me when I did my installs of S10 what was required,
    and I have _never_ installed a prior version of Solaris (or SunOS). (1
    upgrade Ultra10 from S7, 1 install S10 on Duron, 1 update Duron to S10
    U1, 1 install S10 on x86 laptop)

    Okay, I've installed countless other *NIX variants (and non-*NIX OSes)
    over the decades, but it was hardly rocket science!

    Cheers,
    Gary B-)

    --
    __________________________________________________ ____________________________
    Armful of chairs: Something some people would not know
    whether you were up them with or not
    - Barry Humphries

  10. Re: Missing rpc in new install.

    On Sun, 19 Feb 2006 12:41:08 -0700, GreyCloud wrote:

    > Dave Uhring wrote:
    >
    >> On Sat, 18 Feb 2006 21:20:56 -0700, GreyCloud wrote:


    >>>CDE starts, it also shows the first time startup window and the files
    >>>window. But there is also a message window that warns that three files
    >>>need to be checked. The only action is to click on the OK button, which
    >>>leads back to the login screen.

    >>
    >>
    >> If we knew *which* files we could advise which packages need to be
    >> reinstalled. Otherwise it appears that you get to use the Microsfot
    >> solution - completely reinstall the OS taking careful note of any error
    >> messages.
    >>

    >
    > I deleted ~/.dt directory, but it didn't change anything in regards to
    > getting logged into CDE.


    You are not being responsive to the clue. Which three files? You need to
    post the *exact* error message if you expect to get the problem solved.

    >>>Another error inside that file refers to "ld.so.1: dtappgather: fatal:
    >>>/usr/lib/libC.so.5: hardware capability unsupported: 0x1 [FPU]"
    >>>
    >>>Huh? both these libs are 32-bit ELF binaries.

    >>
    >>
    >> The FPU is the floating point processor. What CPU are you using?
    >>

    >
    > A K6 350Mhz. The same one that ran Solaris 8.


    Should be fully supported but perhaps not. I have had no problems with
    either Athlon, Duron or dual Pentium II installations.

    You probably ought to reinstall the OS, selecting the "Complete"
    installation cluster and carefully and completely note any error messages.


+ Reply to Thread