Re: can not allocate log, archival required..

If you are using 8i (or maybe 9i), then as sysdba go into svrmgrl or sqlplus and issue 'archive log start' or change the destination to _1 or _2.

There is a bug in some releases that stops the archiver from noticing it now has space so it has to be told.

Paul Drake <paled_at_comcast.net> wrote in message news:<3C9D62E3.5090403_at_comcast.net>...
> Tina Mourad wrote:
>
> > I get the DOS window warning "Oracle instance xxx can not allocate
> > log, archival required".
> >
> > My init.ora does have the line
> > LOG_ARCHIVE_START = TRUE
> >
> > thanks in advance..
> >
>
> my guess is that you one/many of the following:
>
> 1. no free space in the log_archive_dest location
> 2. very small online redo logs with few groups (little redo capacity)
> 3. online redo, datafiles and archived redo on the same filesystem on a
> single RAID 5 volume.
>
>
> Determine the quantity of redo generated per day during the workday.
> divide that by 8 - that is how large you want your online redo logs to
> be (say 128M for example).
> Create at least 4 online redo log groups of that size:
> - 1 member per group if you have little I/O capacity and non-stringent
> uptime requirements.
> - 2 members per group (on separate physical disks, separate RAID
> controllers or controller channels, at least) if possible
> - 3 members per group if you had the chance to specify the storage
> subsystem.
>
> Your system will have had to have generated 4x the regular hourly rate
> to fill all of the online redo logs.
>
> The archiver process will be able to read from all members in a group in
> a round-robin fashion, thus speeding the archiving of online redo logs.
>
> The only downside is that the size of your cold backup sets will
> increase. If you're only using hot backups, then no matter.
>
> Disks are cheap, Oracle CPU licenses are expensive.
>
> Paul
Received on Tue Mar 26 2002 - 15:23:23 CST