[Glass] System unstable after crash

[Dale Henrichs]

Leo,

You should probably run an object audit to find and repair any repository
level corruption...

It is more likely that you've simply got some application-level corruption
and I would think that you'd have to examine the issues and repair by doing
smalltalk-level surgery on the "damaged" structures.

I'm interested in seeing the stone log to see what error caused the system
crash ...

Dale


On Fri, Jul 18, 2014 at 9:07 AM, Leonardo Andres De Marco <
leo at smalltalking.net> wrote:

> Dale,
> The strange thing is that we try to recover the backup of the day before
> (because users not change any informaton, only use for reporting that day),
> but continue the same error, so, our priority is not the cause of the crash
> (that I assume is because we reach at 100% of the tranlog disk), but to
> restore the system to the user...
>
> You think the log can help in that?
>
> Leo
>
>
> 2014-07-18 11:55 GMT-03:00 Dale Henrichs <dale.henrichs at gemtalksystems.com
> >:
>
> Leo,
>>
>> Most likely the application errors that you are seeing are caused by the
>> missing tranlog records ... the ones that did not get written to disk
>> before the crash ...
>>
>> Could you send a copy of the stone log so that we can see the exact error
>> message that caused the stone to crash?
>>
>> Dale
>>
>>
>> On Fri, Jul 18, 2014 at 7:08 AM, Leonardo Andres De Marco <
>> leo at smalltalking.net> wrote:
>>
>>> Hi!
>>>
>>> Yesterday we have a system crash situation when tranlog disk reach 100%.
>>> That normally happend when the mantainance crontab task have some problem
>>> or when the system day activity is more heavy than expected.
>>>
>>> Anyway, after that situation we execute the mantainance script, delete
>>> old tranlogs files and then restart gemstone and all is good again.
>>>
>>> This time, when we do that we recive some errors (see attachments) and
>>> we fail to make it work, we try also start gemstone in restore mode
>>> (startstone -R) and make a commitRestore (after we delete all tranlogs
>>> files) to close all transactions and syncronize again the extent with the
>>> tranlogs. Again, the operation succeded but when we try to use the system,
>>> we recive the same errors. Also, if you refresh the page, the system is
>>> good again, it seems like a session problem but we not undestand where to
>>> look, some ideas?
>>>
>>> thnks!
>>> Leo
>>>
>>> _______________________________________________
>>> Glass mailing list
>>> Glass at lists.gemtalksystems.com
>>> http://lists.gemtalksystems.com/mailman/listinfo/glass
>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gemtalksystems.com/mailman/private/glass/attachments/20140718/fec11126/attachment.html>