Re: When are redo logs done being archived??

From: Neil Greene <ngreene_at_laoc.SHL.com>
Date: 1996/09/29
Message-ID: <324E215F.586F_at_laoc.SHL.com>#1/1


Bill Beaton wrote:
>
> In article <R.3242C2AF.5F02_at_laoc.shl.com>,
> Neil Greene <ngreene_at_laoc.SHL.com> writes:
> >Mark Jahnz wrote:
> >> What's the best way to determine when archive process has completed the
> >> copy from an online redo log to an archive log. The entries in
> >> v$loghistory appear to show up as soon as the copy is initiated! Does
> >> v$log do the same?
> >Get the list of archive log files (lets call it FLIST), force an archive
> >log switch (alter system switch logfile) and you now know that the file
> >in $FLIST are complete.
>
> Actually, I don't find this absolutely true on systems that do full async
> io. I've found that for our archive logs ... average size 55MB, the
> prompt returns after 'alter systen switch logfile', and the last archive log
> can be as low as 10MB. It can be several minutes before it has been
> entirely flushed from the buffers and caches out to disk.
>
> I've put in a minimum wait 'sleep 1500' after such an operation before I
> feel confident enough to perform an ORACLE shutdown and system reboot.

How can this happen? If I have 10 archive logs, 1-9 which have completed writing and archive log 10 which is currently being written to (in this case your 10Mb archive log) and I force an archive log switch so that I now have archive logs 1-11, then hasn't archive log 10 completed writing?? In which case since my list contains archive logs 1-10. Received on Sun Sep 29 1996 - 00:00:00 CEST

Original text of this message