Details
-
Improvement
-
Status: Open
-
Major
-
Resolution: Unresolved
-
10.2.1.6
-
None
-
Normal
Description
The store locking tests should be run with network server, locking behaviour investigated, and a perhaps a proposal made for testing locking with network server. For network server, the pre-fetching of data causes diffs in the tests and I am not sure if this is expected or not. I noticed for example such diffs in updatelocks.sql and readlocks.sql.
Below is the list of locking tests I see in store. Feel free to take one or two and file a subtask.
EscalateLock.sql
RowLockBasic.sql
RowLockIso.sql
TableLockBasic.sql
TableLockBasic.subsql
lockTableVti.sql
readBtreeCursorLocks.subsql
readBtreeSetLocks.subsql
readCursorLocks.subsql
readSetLocks.subsql
readlocks.sql
updateBtreeCursorLocks.subsql
updateBtreeHoldCursorLocksJDBC30.subsql
updateBtreeSetLocks.subsql
updatecursorlocks.subsql
updateholdcursorlocksJDBC30.subsql
updatelocks.sql
updatelocksJDBC30.sql
updatesetlocks.subsql
RowLockIso.sql
rlliso1multi.sql
rlliso1multi.subsql
rlliso2multi.sql
rlliso3multi.sql
Attachments
Attachments
Issue Links
- is related to
-
DERBY-1342 Run derbyall suite with -Dframework=DerbyNetClient and analyze failures
- Open
Turning this from subtask into independent issue, I think that'll be more useful.
The ultimate goal will be for the tests to get converted to junit and run with network server/client and that any bugs should get logged.
Also note that differences between embedded and network server/client can get linked to
DERBY-310.