In the case where there is a failure to the primary database
serverA and TSA initiates an automated failover to the standby
serverB. Upon restart of the DB2 instance on serverA, the HADR
database is reintegrated as the standby. If a manual takeover is
issued immediately from the new standby on serverA, it may cause
TSA to failback the HADR database to serverB.
This issue is due to a small timing hole in the DB2 automation
logic. It is unlikely to hit this issue. In order to avoid
hitting this issue, it is advisable to wait approximately one
minute before issuing a manual takeover request on the standby
after it has completed automatic reintegration following a DB2
instance level failure.
If this issue is encountered, it should be observed that the
manual takeover request occurred in between the time the
db2V105_start.ksh script was run and the first run of the
hadrV105_monitor.ksh script after instance startup.
For example, in an occurrence of this issue, the manual takeover
occurred at this timestamp:
2015-05-15-10.03.45.004971+480 E1512997E516 LEVEL:
Event
PID : 5564 TID : 140386626430720 PROC :
db2sysc 0
INSTANCE: db2inst1 NODE : 000 DB : HADRDB
APPHDL : 0-9 APPID: *LOCAL.db2inst1.150515020345
AUTHID : DB2INST1 HOSTNAME: HOST01
EDUID : 23 EDUNAME: db2agent (HADRDB) 0
FUNCTION: DB2 UDB, base sys utilities,
sqeDBMgr::StartUsingLocalDatabase, probe:13
START : Received TAKEOVER HADR command.
And from the syslogs, we see the following timestamps for the
db2V105_start.ksh and hadrV105_monitor.ksh script runs:
May 15 10:03:11 HOST01 db2V105_start.ksh[4057]: Entered
/usr/sbin/rsct/sapolicies/db2/db2V105_start.ksh, db2inst1, 0
May 15 10:03:11 HOST01 db2V105_start.ksh[4057]: Able to cd to
/db2/db2home/sqllib :
/usr/sbin/rsct/sapolicies/db2/db2V105_start.ksh, db2inst1, 0
May 15 10:03:11 HOST01 db2V105_start.ksh[4057]: 1 partitions
total: /usr/sbin/rsct/sapolicies/db2/db2V105_start.ksh,
db2inst1, 0
May 15 10:03:17 HOST01 db2V105_start.ksh[4057]: Forcing apps off
before reintegration
May 15 10:03:18 HOST01 db2V105_start.ksh[4057]: After forcing,
starting to reintegrate
May 15 10:03:30 HOST01 db2V105_start.ksh[4057]: Returning 0 from
/usr/sbin/rsct/sapolicies/db2/db2V105_start.ksh ( db2inst1, 0)
First hadr monitor run after reintegration is complete:
May 15 10:03:52 HOST01 hadrV105_monitor.ksh[7234]: Reintegration
memory file created under /tmp for db HADRDB
May 15 10:03:53 HOST01 hadrV105_monitor.ksh[7234]: Must try
reading with db2gcf
May 15 10:03:53 HOST01 hadrV105_monitor.ksh[7234]: auto mode,
and state flags or db2pd fails, use db2gcf: manual:0, usepd:3,
flagFile:0, grc:3
May 15 10:03:57 HOST01 hadrV105_monitor.ksh[7234]: su - db2inst1
-c /db2/db2home/sqllib/bin/db2gcf -t 14 -s -i db2inst1 -i
db2inst1 -h HADRDB returns 1
May 15 10:03:57 HOST01 hadrV105_monitor.ksh[7234]: Returning 2 :
db2inst1 db2inst1 HADRDB |
