db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Knut Anders Hatlen <Knut.Hat...@Sun.COM>
Subject Re: [Proposal for Review] The fix for DERBY-3347 is important enough for the Development Community to issue a recommendation the User Community to upgrade.
Date Tue, 29 Apr 2008 09:07:35 GMT
Stanley Bradbury <Stan.Bradbury@gmail.com> writes:

> Hello Developers -
>
> I would like to follow up on the suggestion made by Kathey Marsden
> that an announcement be posted to derby-user recommending that version
> 10.3 users upgrade to a Derby version that contains the fix for
> DERBY-3347.  The email from Binoy Thomas this morning could be an
> incident of DERBY-3347 [Subject: DB gets corrupts in 10.3.1.2!!]
>
> In hopes of moving this forward I have drafted such an announcement
> and would greatly appreciate review and suggestions from the
> developers most familiar with this bug.  In the message below I have
> listed the reported error and loosely described other errors that
> might be generated should a database become corrupted by this
> defect. I recommend a course of action and use of database consistency
> checks.  If particularly Knut would review and comment on the bug
> related information I would be very grateful.  Word-smithing and other
> adjustments are also welcome.
>
> If this proposal concerns you please respond or I will assume a lazy
> consensus.

Thanks a lot for writing this draft! I think it's a very good idea to
send out a notice, and what you have written looks very good.

> And a couple of ideas to give this proposal more weight:
>
> Would a PMC member volunteer to sponsor this post?  Having an
> acknowledged leader of the development community would give the
> recommendation more weight in many circles. 
>
> Can the recommendation, if posted, be identified as being made by or
> at least supported by the Development Community or a majority of the
> Development Community?  Again this would give more weight and bring
> more attention to the recommendation.  As currently worded the
> recommendation has no author other than the sender of the message.

We could perhaps start a vote to make some kind of official notice on
behalf of the development community and/or the PMC. However, I don't see
a great need for that. Anyone can post a piece of good advice on the
user list. If a user chooses not to listen to it because the person that
posted it isn't regarded as important enough or because it isn't
sanctioned by a high enough authority, that user doesn't have anyone to
blame but himself.

To add some weight, we could refer to the release announcement[1]
(posted by Dyre Tjeldvoll, release manager and PMC member):

  Derby 10.4.1.3 also has many bugfixes, including the fix for
  DERBY-3347 which can cause unrecoverable database corruption.  All
  10.3 users are encouraged to upgrade to avoid potential corruption
  issues.

>  *** For Review and comment  ***
>
> NOTICE TO ALL DERBY v10.3 USERS : CRITICAL FIX NOW AVAILABLE
>
> The Bottom Line:
> It is strongly recommended that you upgrade to Derby 10.4.1.3 to avoid
> any chance of database corruption due to DERBY-3347.  Alternatively
> you can build version 10.3 from the current codeline which also
> contains the fix for this defect.

As others already have mentioned, we should emphasize that the bug only
affects 10.3.1.4 and 10.3.2.1. Perhaps we should even state that users
of Derby 10.2 are not affected?

> This bug can cause unrecoverable database corruption during periods of
> heavy, multi-thread I/O operations.  The error produced in the test
> case used to diagnose the problem was:
>
>  ERROR XSDB3: Container information cannot change once written: was 0,
> now 80
>
> It is felt that other errors might also be generated when this type of
> corruption occurs.  The corruption message will most likely refer to
> page 0 of the container.
>  EXAMPLE:  ERROR XSDG1: Page Page(1039,Container(0, 5856)) could not
> be written...

Since we say that it most likely is page 0, it's better if the example
shows an error message which mentions page 0. Just replace 1039 with 0
in the above message.


[1] http://mail-archives.apache.org/mod_mbox/db-derby-user/200804.mbox/%3cx1o34p9p2e24.fsf@sun.com%3e

-- 
Knut Anders

Mime
View raw message