Message "Fenced server is not null but not connected, Cleanup failed for server" in db2diag.log

Problem(Abstract)

Message "Fenced server is not null but not connected, Cleanup failed for server" in db2diag.log

Symptom

You see a message similar to the following in db2diag.log:

2008-08-11-13.59.07.053103-300 I584734A1271 LEVEL: Error
PID : 581690 TID : 16022 PROC : db2sysc
INSTANCE: my_inst NODE : 000 DB : MYDB
EDUID : 16022 EDUNAME: db2agent (MYDB) 0
FUNCTION: DB2 UDB, Query Gateway, sqlqgRouter_conn_lost_cleanup, probe:25
DATA #1 : <preformatted>
Message: Fenced server is not null but not connected,
Cleanup failed for server: MYSERVER
CALLSTCK:
[0] 0x090000000F632524 sqlqgRouter_conn_lost_cleanup__FP12FencedServer + 0x104
[1] 0x090000000F62DE88 sqlqgRouter__FP17sqlqg_FMP_RequestPP15sqlqg_FMP_ReplyP10sqlri_ufob + 0x348
[2] 0x090000000F633070 sqlqg_fedstp_hook + 0x1B0
[3] 0x090000000AF8988C sqlqgDyload__FP10sqlri_ufob + 0x23C
[4] 0x090000000AF68F34 sqlriFedInvokerTrusted__FP10sqlri_ufobP21sqlriRoutineErrorIntf + 0x1A8
[5] 0x090000000AF68554 sqlriFedInvokeInvoker__FP10sqlri_ufobP14sqlqg_Fmp_Info + 0xE4
[6] 0x0900000009CBA7B0 sqlqg_Call_FMP_Thread__FP17sqlqg_FMP_RequestPP15sqlqg_FMP_Reply + 0x458
[7] 0x090000000B248EE8 sqlqgFedStart__FP14UnfencedServerc + 0x80C
[8] 0x0900000009CC30B0 sqlqgPassthruPrepare__FP15sqlri_rpassthru + 0x2BC
[9] 0x090000000991B5E8 sqlrj_passthru_prepare__FP8sqlrr_cbP11sqlrrstringP14sqlrr_cmpl_envPi + 0x48

Cause

This message can usually be ignored. It can be a side effect of any of the following conditions:

(1) An earlier error occurred that caused Federation Server to lose the connection to a data source. In this case, there will be an earlier error entry in db2diag.log or an earlier error message resulting from an SQL statement that tried to access the data source. A common error message in this case is SQL30081N (SQLCODE -30081).

(2) You have the data source server status health indicator or the nickname status health indicator turned on and Federation Server cannot access the data source. You can determine whether the data source status health indicator or the nickname status health indicator is on by issuing the following commands from the DB2 command line processor (CLP):

GET ALERT CFG FOR DATABASES USING db.fed_servers_op_status;
GET ALERT CFG FOR DATABASES USING db.fed_nicknames_op_status;

(3) Federation Server Version 9.5 or later: You have database configuration parameterauto_runstats turned on for nicknames and Federation Server cannot access the data source, or there is no user mapping from the instance owner to the data source. You can determine whether auto_runstats is on by issuing the following command from the DB2 command line processor and looking at the setting of auto_runstats:

GET DB CFG FOR dbname;

By default, the database configuration parameter auto_runstats is on for nicknames unless there is an auto_runstats maintenance policy that excludes nicknames (for example, specifies TYPE<>'N').

Diagnosing the problem

See the list of causes shown above.

Resolving the problem

This message can usually be ignored. 

If the cause is an earlier error that caused Federation Server to lose the connection to a data source (cause 1 listed above), then resolve the problem that caused Federation Server to lose the connection.

If the cause is a data source server status health indicator or a nickname status health indicator (cause 2 listed above), then resolve the problem that prevents Federation Server from accessing the data source, or turn off the health indicators. To turn off the health indicators, issue the following commands from the DB2 command line processor:

UPDATE ALERT CFG FOR DATABASES USING db.fed_nicknames_op_status SET THRESHOLDSCHECKED NO;
UPDATE ALERT CFG FOR DATABASES USING db.fed_servers_op_status SET THRESHOLDSCHECKED NO;

Federation Server Version 9.5 or later: If the cause is auto_runstats being on (cause 3 listed above), then do one of the following, listed in order of preference:

• Resolve the problem that prevents Federation Server from accessing the data source. In particular, make sure there is a user mapping from the instance ID to the data source.
• Create an auto_runstats maintenance policy that excludes nicknames belonging to servers that Federation Server cannot access.
• Turn off auto_runstats for the database. To turn off auto_runstats, issue the following command from the DB2 command line processor:

UPDATE DB CFG FOR dbname USING auto_runstats OFF;

Note that this turns off auto_runstats for all tables and nicknames, including local tables. DB2 will no longer maintain table or nickname statistics automatically, which can result in performance degradation.

https://www-304.ibm.com/support/docview.wss?uid=swg21318816