• Bug#265996: debian-installer report

    From Ben Burton@1:229/2 to All on Mon Aug 16 10:00:09 2004
    XPost: linux.debian.maint.boot
    From: [email protected]

    Package: installation-reports

    INSTALL REPORT

    Debian-installer-version: rc1, downloaded from ftp.au.debian.org.
    uname -a: Linux metebelis 2.4.26-r4k-ip22 #1 Sun Jul 18 20:49:31 UTC 2004
    mips GNU/Linux
    Date: Sun Aug 15, 5pm GMT+10
    Method: Installed over a serial console (no keyboard or monitor) using
    bootp/tftp. Serial cable from the mips box was connected to an
    i386 machine running minicom, where i did the actual typing.
    dhcp server and tftp server were on a third box (i386/sid), all
    on a local subnet. Debian packages were downloaded from
    ftp.au.debian.org, no proxy used.

    Machine: SGI Indy
    Processor: R4600, 133MHz
    Memory: 64M
    Root Device: SCSI (/dev/sda)
    Root Size/partition table: 1.06G in size.
    Approx. 16M for SGI volhdr (part. 9),
    then approx. 590M for / (ext3) (part. 1),
    then approx. 418M swap (part. 2).
    Output of lspci and lspci -n:
    metebelis:~# lspci
    pcilib: Cannot open /proc/bus/pci
    lspci: Cannot find any working access method.
    metebelis:~# lspci -n
    pcilib: Cannot open /proc/bus/pci
    lspci: Cannot find any working access method.
    metebelis:~#

    Base System Installation Checklist:
    [O] = OK, [E] = Error (please elaborate below), [ ] = didn't try it

    Initial boot worked: [O]
    Configure network HW: [O]
    Config network: [O]
    Detect CD: [ ]
    Load installer modules: [O]
    Detect hard drives: [O]
    Partition hard drives: [*] (worked with difficulties)
    Create file systems: [*] (worked with difficulties)
    Mount partitions: [O]
    Install base system: [O]
    Install boot loader: [O]
    Reboot: [O]

    Comments/Problems:

    Mostly went very smoothly -- I was quite pleasantly surprised I could
    do the entire installation over a serial console without a monitor or
    keyboard.

    Biggest difficulty was partitioning. The machine originally came with irix installed, and the machine already had four partitions (1:boot, 2:swap, 9:volhdr and 11:volume(iirc)).

    I deleted partitions 1, 2 and 9 and recreated them with different sizes.
    I then went on to the next d-i stage (create filesystems), and I was given an error that there were no partitions available for creating filesystems on
    (iirc -- alas I didn't write this down). I went back to the partitioner and deleted all four partitions (1,2,9,11). I recreated 1, 2 and 9 and found
    that 11 had automagically recreated itself (though with 1010 sectors instead
    of the original 1011 from the initial irix partitioning). d-i was then quite happy to carry on from here.

    One problem I did have with the partitioner was that you couldn't read the
    full partition table on an 80-column terminal. The device names were
    so long (/dev/scsi/host0/bus0/target1/lun0/part1, etc.) that the table
    was pushed too far to the right -- the partition size was obfuscated,
    and anything to the right of it (such as the partition type) could not be
    read at all.

    The other trouble I had was during package configuration over the serial console, in that I could not always see what options I was choosing
    during the base system configuration.

    Specifically, when running minicom from a konsole under X, the bold text
    looked identical to standard text; as a result I was working blind in the
    list selections (e.g., which timezone are you in, how do you want
    exim to handle your mail). When running minicom from tty1 I could handle
    the lists fine (since the bold showed up), but some yes/no choices were
    unclear (e.g., does your hardward clock use GMT) -- there was no indication
    of which option i currently had selected. In both the konsole and tty1 scenarios, the list of tasks (devel, desktop, manual package selection, etc) gave no indication as to where my cursor was -- I had to press space at
    random intervals to see which option changed state so i could navigate.

    Perhaps these navigation issues could be improved by using a visual feedback that does not rely on cursor position and font attributes alone? An example might be an asterisk that moves up and down a list (or between two buttons)
    as you press the arrow keys.

    All in all though, the installation was a very pleasant experience. My congratulations to the d-i team for such a smooth process.

    Ben.


    --
    To UNSUBSCRIBE, email to [email protected]
    with a subject of "unsubscribe". Trouble? Contact [email protected]

    --- SoupGate-Win32 v1.05
    * Origin: you cannot sedate... all the things you hate (1:229/2)