By using this site, you agree to our updated Privacy Policy and our Terms of Use. Manage your Cookies Settings.
448,781 Members | 1,387 Online
Bytes IT Community
+ Ask a Question
Need help? Post your question and get tips & solutions from a community of 448,781 IT Pros & Developers. It's quick & easy.

V9.5 HADR On AIX

P: n/a


Hi,

Testing the above without TSA or HA, just plain HADR performing manual
db2 TAKEOVER HADR ......................... etc.

I am testing without the PEER_WINDOW i.e. set to 0 and HADR_SYNCMODE
= SYNC.

A quick question, say the primary and standby database servers are in
PEER state and the primary database server suddenly crashes through a
panic etc. What is the procedure to determine if forced takeover on
the standby database server can be issed without the loss of data.

Is there a set of rules or procedure?

Many Thanks.
Oct 29 '08 #1
Share this Question
Share on Google+
1 Reply


P: n/a
<ag********@hotmail.comwrote in message
news:ff**********************************@u57g2000 hsf.googlegroups.com...
Hi,

Testing the above without TSA or HA, just plain HADR performing manual
db2 TAKEOVER HADR ......................... etc.

I am testing without the PEER_WINDOW i.e. set to 0 and HADR_SYNCMODE
= SYNC.

A quick question, say the primary and standby database servers are in
PEER state and the primary database server suddenly crashes through a
panic etc. What is the procedure to determine if forced takeover on
the standby database server can be issed without the loss of data.

Is there a set of rules or procedure?

Many Thanks.
There will be no loss of data, because the standby server will rollback the
uncommitted transactions, just as will the primary server when it is brought
back up and goes into crash recovery.

You will loose data that has not be been committed on the primary server,
but that has nothing to do with HADR and is normal for all modern database
systems.

BTW, you might want to consider using Nearsynch instead of Synch mode. The
only difference is that in synch mode DB2 guarantees that the logs have been
sent to the standby and written to the HADR buffer and also written to disk
on the standby when a transaction is committed; whereas with Nearsynch DB2
guarantees that the logs have been written to the HADR buffer on the standby
(which eventually will get written to disk). It is inconceivable to me how
one could loose any data with nearsynch unless that standby server crashed a
few milliseconds after the primary server crashed. If that happened (both
servers crashed at the same time) you would not do a takeover to the standby
server (since it would also be down), so you could not loose any data, and
when you brought the primary and standby server back up, the logs would get
in synch.
Oct 29 '08 #2

This discussion thread is closed

Replies have been disabled for this discussion.