You didn't mention under what conditions you want the answer, but...
There are at least the following situations where some part of HADR
may be looking for an existing log file.
Scenarios:
(1) HADR Standby needs to read a log file during "local
catchup" (initial processing similar to crash recovery, if needed, and
beginning rollforward).
(2) HADR Primary needs to ****p log data from a log file to HADR
Standby while the standby is performing "remote catchup" (after local
catchup; standby continues to catch up to primary using log data from
older log files not found on standby but that can be retrieved at
primary).
I would think that in case (1) the log file in overflowpath should be
found (at least if a copy of same named file does not already exist in
the active log path). The HADR standby behaves much like a
rollforward process, and overflowlogpath is one place rollforward
should look for a log file. In fact, either rollforward or HADR could
potentially restore a log file to the overflowlogpath.
In case (2), I'm not as certain. When the primary needs to find an
old log file to ****p to the standby, it is not in a rollforward type
environment. I don't know if it would find a log file previously
placed in overflowlogpath, though that would be nice :-) behavior. If
the primary must restore an old log file from archive in order to ****p
it to the standby, it could presumably end up using overflowlogpath as
the target for that restore; in that case it should of course find the
file. So my guess (not verified by experience or experimentation) is
more likely than not files in overflowlogpath would be seen in this
case as well -- likely the common code to find a log file to read
looks in all the usual places no matter where in DB2 it is invoked.
Regards,
- Steve P.
--
Steve Pearson, DB2 for Linux, UNIX, and Windows, IBM Software Group
"****tland" Development Team, IBM Beaverton Lab, Beaverton, OR, USA
|
