Knut Anders Hatlen (JIRA) wrote:
[ http://issues.apache.org/jira/browse/DERBY-1248?page=comments#action_12419280 ]
Knut Anders Hatlen commented on DERBY-1248:
-------------------------------------------
I also see this failure on Linux, but not as frequently as on Solaris (had to
run the test 125 times on Linux, whereas it fails after about 25 runs on
Solaris). I have not been able to reproduce the failure on machines with single
CPU.
The existing attached zip does not have the database and logs, so not
much help except for the stack and error message. Posting a few of
your failures may help a lot. If you can't post complete directories,
derby.log, data/log - can you let us know if the error is always the
same (ie. row offset 140 vs 2? or maybe different values).
Usually with errors during recovery boot phase if we can get the data
and the log it will reproduce and we can get a log of info as booting
again will follow the exact same logic. Unfortunately the actual bug
probably happened when the logs were written rather a bug in recovery.
The fact that you can reproduce is important, I would now consider this
more important to fix for 10.2.
