Import a CSV file into a PostgreSQL database, then in the schema browser list the sessions and you will see a bunch of new sessions.
|
120 KB
ImportThread.java Line 420 I added debug code :
String sessionID = ConnectionUtil.getResultData(conn, "SELECT pg_backend_pid()", 1); System.out.println("Session: " + sessionID);
The session ID is the same for each record insert.
We are using the ClusterDBConnection.java for PostgreSQL because PG doesn't allow database context which. If I debug an import I can see the ClusterDBConnection._map only has two connections during the entire import.
Putting a breakpoint in ConnectionProperties.getConnection(x,x,x) line 1976 you can see the stacktrace and see that the new connection call is being called because of a call from ConnectionProperties.requiresEscapeBackslash(). The problem is caused by a change made by Ivan because of issue #13085. Revision #42391.
I have made changes to connectionProperties.requiresEscapeBackslash() method so that no connection will ever be made here. The connection will need to passed through to the method. If the connection is null the result will be fixed to true for postgres 9+ else false. This modification to the method required a bit of refactoring and some code shifting so that we only check this once for the connection instead of everytime we try to execture a transaction in some areas. The extent of the code changes affect aquaactivate, aquaclusters, datastudio & the aqua common server classes. I have run through testing data studio areas but will need help testing some of the areas. Currently I've only made the code changes to trunk...
I have made changes to connectionProperties.requiresEscapeBackslash() method so that no connection will ever be made here. The connection will need to passed through to the method. If the connection is null the result will be fixed to true for postgres 9+ else false. This modification to the method required a bit of refactoring and some code shifting so that we only check this once for the connection instead of everytime we try to execture a transaction in some areas. The extent of the code changes affect aquaactivate, aquaclusters, datastudio & the aqua common server classes. I have run through testing data studio areas but will need help testing some of the areas. Currently I've only made the code changes to trunk...
After fixing many issues I caused refactoring this code, I think the code looks stable. I have made the needed changes to back port to 15 once we have tested and are good with the current code in 16.
After fixing many issues I caused refactoring this code, I think the code looks stable. I have made the needed changes to back port to 15 once we have tested and are good with the current code in 16.
Verified in AD Server and no new connections are being opened up. Tests were done for PostgreSQL only
Verified in AD Server and no new connections are being opened up. Tests were done for PostgreSQL only
Issue #13116 |
Closed |
Fixed |
Resolved |
Completion |
No due date |
Fixed Build v16.0.0-beta-44 |
No time estimate |
2 issue links |
is broken by #13085
Issue #13085Table Data Editor on PostgreSQL should check "show standard_conforming_strings" in ConnectionProperties.requiresEscapeBackslash() |
relates to #13085
Issue #13085Table Data Editor on PostgreSQL should check "show standard_conforming_strings" in ConnectionProperties.requiresEscapeBackslash() |
ImportThread.java Line 420 I added debug code :
The session ID is the same for each record insert.
We are using the ClusterDBConnection.java for PostgreSQL because PG doesn't allow database context which. If I debug an import I can see the ClusterDBConnection._map only has two connections during the entire import.
Putting a breakpoint in ConnectionProperties.getConnection(x,x,x) line 1976 you can see the stacktrace and see that the new connection call is being called because of a call from ConnectionProperties.requiresEscapeBackslash(). The problem is caused by a change made by Ivan because of issue #13085. Revision #42391.