Re: Archive Processes

Mladen Gogala wrote:
> On Thu, 01 Sep 2005 14:49:49 -0700, Joel Garry wrote:
>
> > Your problem is fixed in 10g. See
> > http://metalink.oracle.com/metalink/plsql/ml2_documents.showFrameDocument?p_database_id=BUG&p_id=3142091
> > The essential problem is that no arch process is dedicated local -
> > Oracle starts up as many arch processes as it needs, but eventually they
> > all get stuck waiting for the network.
>
> Weeeeelllll, there is an unsupported, undocumented solution which works
> with 9.2 as well.... Of course, it involves an underscore parameter:
> _LOG_ARCHIVE_CALLOUT = 'LOCAL_FIRST=TRUE'
> If the above parameter is set then the ARCH process will begin archiving
> to the local destination first. Once the redo log has been completely and
> successfully archived to at least one local destination, it will then be
> transmitted to the remote destination.

Thanks Mladen, I had forgotten about that, though it was trying to bubble up through my subconscious, hence my suggestion to call support and ask about options. Not totally undocumented, see Note: 260040.1 . Not a big fan of newly introduced backported parameters, personally.

Of course, since the OP has set that and still has the problem... (Sushil, perchance do you work for WAMU?) I would guess that the effects of the parameter are hidden due to there being local disk saturation, that is, local archiving is overwhelmed by the rapid log switches. Still need more, bigger redos, probably some redo log buffer tweaking, more network bandwidth and wonders if there are disk layout/contention issues for both the redo and arcs.

jg

--
@home.com is bogus.
http://www-personal.umich.edu/~mejn/election/countymaplinearlarge.png