On Fri, Mar 10, 2006 at 11:53:15AM +0200, Denis Chapligin wrote:
> And now it doesn't starts with following info:

Too bad search on www.coda seems to be broken right now, because I'm
pretty sure I've seen this before.

> Date: Fri 03/10/2006
> 10:38:36 Coda Vice, version 6.0.14  log started	at Fri Mar 10 10:38:36 2006
> 10:38:36 Main process doing a LWP_Init()
> 10:38:36 Main thread just did a RVM_SET_THREAD_DATA
> 10:38:36 Setting Rvm Truncate threshhold to 5.
> log_recover failed.
> do_rvm_options failed
> 10:38:36 rvm_init failed RVM_EIO

Somehow the RVM log must have gotten corrupted. Luckily Coda server tend
to flush pretty frequently, so there is a 99% chance that everything
will work when we reintialize the log. But just in case,

Could you get the rvm parameters out of /etc/coda/server.conf, i.e.

    rvm_log="..."
    rvm_data="..."
    rvm_data_length="...."

And then run

    rvmutl $rvm_log
    * status
    Status of log:	   /vice/LOG
    log created on:        Tue Apr 13 2004 23:29:10.485906
    log created with:      RVM Interface Version 1.3  7 Mar 1994
			   RVM Log Version  1.4 Oct 17, 1997
			   RVM Statistics Version 1.1 8 Dec 1992
    status last written:   Thu Jan  6 2005 19:05:39.203547
    last truncation:       Thu Jan  6 2005 19:05:39.203545
    log head offset:            80436
    log tail offset:            80436
    log empty:			 true
    space used by records:          0
    space available:          2097152
    status area size:            1536
    total log size:           2098688
    first record number:            0
    last record number:            77

This should give some idea of how dirty your log actually is.

rvmutl has a bunch of other commands, one of which is 'recover' which is
what the server uses during startup. It can also be used to reinitialize
and clear the existing log.

Jan