SQL connections not disconnecting - new connections refused

jsmithqbp

New Member
I've set up a secondary broker for SQL connections. I'll likely be increasing the # of sql server connections, but this is what we started out with.
[servergroup.nxt.defaultconfiguration.secondaryservergroup]
configuration=nxt.defaultconfiguration
displayname=secondaryServerGroup
maxclientsperserver=1
minclientsperserver=1
numberofservers=5
port=nxtsql
type=sql



We have connections showing in Promon, though the users say they are not actively connected. This is causing new connections to be refused: OE Broker fails to find server available for connecting.

Usr Name Type Wait Trans id Login time Serv

411 REMC -- 0 0 07/12/11 10:22 4
412 REMC -- 0 0 07/10/11 06:20 6
413 reports REMC -- 0 1171881698 07/12/11 08:43 9
416 REMC -- 4176 0 07/12/11 13:04 4
417 REMC -- 0 0 07/10/11 06:19 4
418 REMC -- 0 0 07/15/11 10:29 2
419 ecomm REMC -- 164800 0 07/06/11 14:23 3
420 REMC -- 0 0 07/11/11 12:01 2
422 ecomm REMC -- 149200 0 07/07/11 10:19 7
423 REMC -- 0 0 07/15/11 02:00 2
425 reports REMC -- 0 1173650701 07/13/11 08:02 8


It seems the connections are not disconnecting as they should. But I'm not sure why. If I shut down one of the users, then I'm able to connect. In the above snapshot, "reports" is connecting via ODBC; "ecomm" via JDBC.

Why wouldn't the connections be cleanly disconnecting? I read something about being able to disconnect idle 4GL users, can I do the same for SQL users? Should I have to or does it sound like something's wrong?
Our database log files are

Progress (AIX): 10.1B03
JDBC: 10.1
ODBC: 10.1B 32bit
 
I would say that there is something going on that is wrong. But I doubt that the problem is on the database side. Rather I would say that the ODBC/JDBC clients do not disconnect gracefully thus leaving the database thinking that they are still connected. No matter what the users say you should check how they connect to the database and what they do when they think that they "disconnect" from the database. Probably it is just a setting on the driver that could help but for that I am no expert.

Heavy Regards, RealHeavyDude.
 
Here's what we've changed which seems to have resolved the issue.

  • Changed the default isolation level to “Read Uncommitted”.
  • Changed the “Fetch Array Size” to 25, down from the default of 50.
 
Back
Top