RE: [NeoStats-Devel] GDBM problem - WAS Re: [NeoStats-Devel] NeoStats 3.0 alpha 2 pre release notice

> DeadNotBuried wrote:
> > Mark Wrote:
> > > I am not able to reproduce this so will need a debug.log to 
> > look into 
> > > it further.
> > 
> > I've got a debug.log (same address as usual to grab it). 
> > things seem to have changed slightly although not sure why. 
> > when unloading and loading a module, everything does get set 
> > back to defaults, but on shutting down and restarting 
> > NeoStats all settings come back to what they were before, not 
> > as default.
> > 
> > I did a set list on Secureserv, unloaded it, loaded it, did a 
> > set list again. shut down NeoStats, restarted NeoStats and 
> > did a set list again on secureserv.
> 
> The debug log shows an error:
> 
> 09/03/2005[01:09:09] DEBUG4 SecureServ - DBMOpenTable
> 09/03/2005[01:09:09] DEBUG1 SecureServ - gdbm_open fail: 
> Can't be writer
> 

Can we bail on errors like this? (as in Exit NeoStats....) I'm sure that
there will be many users that will do something funky, and we will get
errors like this in the log files, and users being users, they will not
check log files, and the log support requests because their settings or data
don't get saved. 




> This explains the problem you are seing. AFAICT, this is a 
> locking issue and
> the only thing that can really explain why a shutdown works OK is the
> timespan. So, a couple tests for you to try. 
> 
> 1) Try the same test you did for the latest debug.log on 
> NeoStats rev 2332,
> i.e. unload module then load immediately. This revision removes the
> superfluous sync that was added recently to the close table 
> call. It may or
> may not have an effect but the sync should not be there and 
> may or may not
> have contributed to the problem.
> 
> 2) Try a delay between unload and load module of say a minute 
> or two. This
> should give time for any cache to clear. There should not be any cache
> active during this process but this will check things a 
> little and see if
> time does affect the issue and therefore explain any intermittency in
> reproduction.
> 
> This is also something that any other NeoStats 3.0 SVN users 
> can try to
> determine whether this is specific to BSD or occurs on other operating
> systems.
> 
> Mark.
> 
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: devel-unsubscribe-wool9L35kifE9wlyV4mCnKxOck334EZe@public.gmane.org
> For additional commands, e-mail: [email protected]
> 
>