Kathey Marsden wrote:
Rick Hillegas wrote:
I think it's worth spinning a new release candidate for this problem.
We might want to wait another day to see if people trip across other
serious problems with the current RC.
I think we should wait more than a day. I think it would be good to
get the first round of Buddy testing done. I think I should be able
to finish update statistics by Friday. Also we still have DERBY-4075
to contend with. I plan to vote against the release if this issue is
outstanding.
If you plan to vote against the release if DERBY-4075 is outstanding, I
think we need to lay a plan for it.
I've been running the test for weeks without seeing the issue. The
largest database grew to 67 GB (plus 67 GB for the backup), another one
grew to 32 GB.
If we clearly communicate the environment where the problems have been
seen, people would get the chance to start test runs of their own.
After all, starting the test itself is very simple.
I have also seen an issue with multi-threaded inserts (25 threads
against the same table) and a subsequent table compress, but I have to
investigate further (I was unable to recover the stack trace, I just got
the error message).
The table grew to around 20 GB, but the amount of data in it should be
far less (maybe around 2 GB, ten times less). When compressing, Derby
threw an IllegalArgumentException, saying "Illegal Capacity: XXX". This
can look as an overflow or a calculation error when sizing one of
Vector, ArrayList, or Hashtable.
I'll try to reproduce and get more information.
--
Kristian
Speaking of buddy testing, we still have some features unassigned.
http://wiki.apache.org/db-derby/TenFiveOneBuddyTesting
Students, buddy testing is a great way to start contributing to the
community and requires little set up. You just need to update the
Wiki, download the release candidate and do some ad hoc testing of
the feature and documentation.
Kathey
