AIX 5.3, DB2 UDB v 8.2

I was curious if there was a mechanism in HADR to switch the
HADR_REMOTE_HOST. If a link or switch fails and there is an alternate
path is there a way to use the path?

I tried changing the remote host on the machines on the fly but I
noticed that it states the changes do not take affect right away.
Then the Standby rebooted. I did finally get both machines up with
the new port configuration but they wouldnt talk to each other. The
Standby logged that there was no Primary connected and the Primary
timedout as it did not hear from the Standby. So I take it you cannot
change the remote hosts this way.

Any other possiblities? I would like to be able to fail over to
another path if possible.

Thanks.

HADR config parameter updates are not dynamic; the databases must be
bounced for the changes to be taken up.

>From your description my only guess as to what happened that no
connection was made after the configuration changes is that perhaps
you only updated one side. Either the primary was not listening on
the host/port that the standby tried to connect to, or there was an IP
address mismatch.

HADR presently requires that the local and remote host configurations
resolve to the same IP addresses on both sides. If not, HADR won't
make a connection. This has been raised as an issue for certain
environments such as where a NAT sits between the primary and
standby. We expect to provide a means to relax the sanity check in a
future release.

Regards,
- Steve P.
--
Steve Pearson, DB2 for Linux, UNIX, and Windows, IBM Software Group
"Portland" Development Team, IBM Beaverton Lab, Beaverton, OR, USA

"Steve Pearson (news only)" <stevep222@my-deja.com> wrote in message
news:1170894801.455531.121510@k78g2000cwa.googlegr oups.com...
> HADR config parameter updates are not dynamic; the databases must be
> bounced for the changes to be taken up.
> Regards,
> - Steve P.

What is definition of database bounced? No connections to the database (not
activated), or restart of the instance?

On Feb 7, 8:59 pm, "Mark A" <nob...@nowhere.com> wrote:
> "Steve Pearson (news only)" <stevep...@my-deja.com> wrote in messagenews:1170894801.455531.121510@k78g2000cwa.g ooglegroups.com...
>
> > HADR config parameter updates are not dynamic; the databases must be
> > bounced for the changes to be taken up.
> > Regards,
> > - Steve P.
>
> What is definition of database bounced? No connections to the database (not
> activated), or restart of the instance?

I was curious of this too...Is there an official procedure for this to
make it right in the db2 world? I am going to do some additional
testing but on my original test I changed the remote_hosts on both
side but after I read Steve's response I realized I must have to also
change the local_host on both sides??.

> What is definition of database bounced? No connections to the database (not
> activated), or restart of the instance?

I meant shutting down and restarting the database. Since HADR
explicitly activates the database, the usual approach would be to
issue "deactivate db" + "activate db" commands.

Regards,
- Steve P.
--
Steve Pearson, DB2 for Linux, UNIX, and Windows, IBM Software Group
"Portland" Development Team, IBM Beaverton Lab, Beaverton, OR, USA

> I am going to do some additional
> testing but on my original test I changed the remote_hosts on both
> side but after I read Steve's response I realized I must have to also
> change the local_host on both sides??.

Yes, if the actual host name changes, one side's hadr_local_host must
change, as well as the other side's hadr_remote_host...unless the
change is hidden (for example, if the setup is configured by ip
address instead of host name, and the ip address didn't change).

Regards,
- Steve P.
--
Steve Pearson, DB2 for Linux, UNIX, and Windows, IBM Software Group
"Portland" Development Team, IBM Beaverton Lab, Beaverton, OR, USA