how to disable HADR in db2

ravichandrapratap · #1 (**permalink**) 01-27-12, 07:24

Hi all,
we had HADR setup on windows, due to some reasons i want to down the standby server , at the same time i want to run primary server.
can anybody provide me procedure to achieve this
Any help can be appreciable..

Marcus_A · #2 (**permalink**) 01-27-12, 07:28

deactivate the database on the standby server
stop hadr on the standby server
stop hadr on the primary server

See the manual for exact command syntax.

ravichandrapratap · #3 (**permalink**) 01-30-12, 04:40

Quote:

Originally Posted by Marcus_A

deactivate the database on the standby server
stop hadr on the standby server
stop hadr on the primary server

See the manual for exact command syntax.

Thanku so much Marcus, i tried above commands only, now everything is fine,

ravichandrapratap · #4 (**permalink**) 01-30-12, 04:46

As i am new to HARD,
i have following doubts regarding this

1q) After few days i have to enable HADR again, for this one simply restarting HADR on both the servers is enough or
2q)i have to take backup on primary and restore on standby server
3q)if i have to take backup , it must be online or offline??

przytula_guy · #5 (**permalink**) 01-30-12, 05:37

if the logfiles are still available on primary (archived to disk-tape-tsm) and can be retrieved - you can start hadr on stdby - will do remote catchup and sync
if not
take online backup on primary and restore on stndby - NO ROLLFORWARD todo
db will sync...

sathyaram_s · #6 (**permalink**) 01-30-12, 05:40

In theory, you should be able to start HADR without the need for restore[ assuming - you have not rolled forward the standby ] . But this will mean the standby has to catchup with the primary - with the primary doing the work of retrieving the log files and sending across the network. Depending on how much the primary has changed, this process will take a while.

So, if the database is small, it is easier to do a backup and restore. The backup can be either online or offline.

Alternatively, you can use do "manual" catchup. Just copy the archive logs generated on the primary and issue a ROLLFORWARD command on the standby . Make sure you do not COMPLETE the rollforward. Once the databases are reasonably in sync, then start HADR. The catchup will be quicker.

Cheers
Sathy

db2mtrk · #7 (**permalink**) 02-01-12, 15:11

hi,
you might find the answer by this time.
However,
Once you use the STOP HADR command on the standby , the db becomes online. which means you can not restart HADR without restoring it.

Never use the STOP HADR command unless you are ready to restore the DB.

mtrk.

sathyaram_s · #8 (**permalink**) 02-01-12, 15:27

As far as I rememer, STOP HADR puts the db on standby in rollforward pending state. It requires a rollforward command to bring it online.

As long as the db is in rollforward pending state, it retains all its internal identifiers and therefore can get in sync with the primary if HADR is started.

Cheers
Sathyaram

db2mtrk · #9 (**permalink**) 02-01-12, 16:05

Quote:

Originally Posted by sathyaram_s

As far as I rememer, STOP HADR puts the db on standby in rollforward pending state. It requires a rollforward command to bring it online.

As long as the db is in rollforward pending state, it retains all its internal identifiers and therefore can get in sync with the primary if HADR is started.

Cheers
Sathyaram

Yes, you are right, the SB will be in rollforward pending state (sorry not online). but you can not start the HADR. I have tried this many times. Finally I had to restore it.

Regards, mtrk

sathyaram_s · #10 (**permalink**) 02-01-12, 16:43

Guy seems to concur with my (above) opinion this thread.

If I remember correctly, I have done the "manual" catch-up I referred in my earlier post when the standby server had crashed and was lagging behind the primary by dozens of logs when it became available.

I could be wrong though. I do not have access to a HADR pair to test.

przytula_guy · #11 (**permalink**) 02-02-12, 02:15

you can stop hadr as much as you need and sometimes it gets stopped (by bug or network problem) as long as the log files are available somewhere - restore is not needed or should not be..

ravichandrapratap · #12 (**permalink**) 02-02-12, 06:39

Thanku so much guys for your response, sorry for the late reply...

satyaram_s::Alternatively, you can use do "manual" catchup. Just copy the archive logs generated on the primary and issue a ROLLFORWARD command on the standby . Make sure you do not COMPLETE the rollforward??

Pls can you explain me in detail , which rollforward command i have to prefer for this activity.

przytula_guy · #13 (**permalink**) 02-02-12, 08:04

there is just 1 command : rollforward
see for details : ROLLFORWARD DATABASE
DO NOT specify AND COMPLETE

ravichandrapratap · #14 (**permalink**) 02-02-12, 09:02

Thanku so much Przytula for your reply,
Can anybody explain difference between HADR_SYNCMODE=SYNC and
HADR_SYNCMODE=NEARSYNC, i already refered to some basic definition of both from IBM site , but i am not getting clear idea.

For my situation (i.e , to stop DR server and after few days again i have to enable HADR)
which one is more suitable(whether HADR_SYNCMODE=SYNC or
HADR_SYNCMODE=NEARSYNC)