[9fans] venti: decompression gave partial block

[9fans] venti: decompression gave partial block

[Michael Teichgräber]

Hi,

I am getting the above message when I try to boot my fossil/venti
system, after entering the user name. The detailed message is

    time...
    venti...err 2: decompression gave partial block: 5433/8192

    clump=48115 failed to read correctly: d-g-p-b: 5433/8192

    fatal /boot/venti error: can't sync server: d-g-p-b: 5433/8192

    fossil(#S/...)...warning: connecting to venti: connection refused
    version...time...
    boot: /386/init: venti i/o error block f1dfa...faaf: no connected to
    venti server
    panic: boot process died: unknown
    panic: boot process died: unknown
    dumpstack disabled
    cpu0: exiting


Before this situation occured I was trying to boot the system with a
minimal modified 9pcf kernel (one extra quotefmtinstall() call), and
needed some reboots until I got my plan9.ini right. Within the last
step I bootet the system up, did

  % 9fat:
  % sam /n/9fat/plan9.ini
  % unmount /n/9fat
  % fshalt

and pressed ^T^T^r. This sequence didn't take longer than one, two
minutes. I realized some slight hard disk access, which seems to be
explained by http://9fans.net/archive/2006/03/487. I could imagine
that this last reboot somehow interrupted a venti activity.

Do you know something I could do at the 9load prompt (or out of the CD
live system) to help venti recover? Perhaps venti could be made forget
about that faulty block?

Thanks in advance,

Michael

Re: [9fans] venti: decompression gave partial block

[erik quanstrom]

you can boot the livecd and mount your harddrive.  perhaps your
kernel or plan9.ini is the source of the problem.

- erik

Re: [9fans] venti: decompression gave partial block

[Michael Teichgräber]

erik quanstrom <quanstro@quanstro.net> writes:

> you can boot the livecd and mount your harddrive.  perhaps your
> kernel or plan9.ini is the source of the problem.

It's running again! Booting the livecd and running
    venti/checkarenas -f /dev/sdC0/arenas.

resulted in a message about having fixed some header fields of the block
in question.

Using -af options, and then -a still shows that partial block message,
but venti seems to get along with it now (perhaps it is sufficient if
the end section of an arena has been fixed).

Michael