tech-install archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Should build.sh install=/foo generate a "complete" installation?



I’ve just used “build.sh install” for the first time in quite a while—the last time was when 9.0 was current—to set up an NFS root directory for a 64-bit SPARC system. I remember it working in the past to set up a full-but-not-yet-configured installation of NetBSD; today, however, it didn’t put things in place like the default /etc/passwd, /etc/groups, or /etc/rc.conf so actually setting up and configuring the system from a newly-built NetBSD seems like it’ll be difficult.

Specifically, I used

    ./build/sh \
        -m sparc64 \
        -T /usr/tools-10/sparc64 \
        -O /usr/obj-10/sparc64 \
        -x -U -u \
        ${command}

three times to build (my local branch of) netbsd-10, with ${command} producing in order:

    tools
    kernel=GENERIC distribution
    install=/export/NetBSD/10.1/root/ultra60

to build tools for sparc64, to build a kernel for sparc64, and then to install it to the NFS-exported root directory that I want to use for my Sun Ultra 60.

Upon doing the install to an empty directory, I get failures in the following post-install checks:

  atf autofsconfig bluetooth ddbonpanic defaults dhcpcd envsys fontconfig
  gid gpio iscsi makedev mtree named opensslcertsconf opensslcertsrehash
  pam periodic pf pwd_mkdb rc ssh uid

This… doesn’t seem right? Is there a different way that I should set up such a directory? I don’t want to go through sysinst for every single system.

  -- Chris



Home | Main Index | Thread Index | Old Index