Re: Context Option & REDO Logs

On 7 Oct 1998, TimKArnold wrote:
>
> Also, running Oracle Context Option 2.3.6.0.0. After starting the OCO
> threads or personalities (QDM) and Context Indexing is started, there is
> is often a interrupt from Oracle80.exe
> 'Warning - The following error occurred during Oracle redo log archival
> Oracle instance XXXX - Cannot allocate log, archival is required
> Press <ENTER> to acknowledge message'

You are generating redo during faster ConText indexing than the archiver can archive. When all the online redologs get filled up, the archiver is still busy archiving previous log groups.

There are several options that you may want to pursue.

You could increase the efficiency of the archiver by having it archive to a faster disk subsystem.

You could increase the number of online redolog groups to such a size that your indexing job could finish without all online redolog groups remaining unarchived. This probably would require an unreasonably large number of online redolog groups.

The best solution is probably be to try to reduce the generation of redo by using the UNRECOVERABLE option for that portion of the ConText indexing routine that actually generates index segments in the ConText data dictionary. Unfortunately, I don't think ConText provides a facility to do this. It would, of course, mean rebuilding those indexes in the event of a recovery.

--
Jeremiah Wilton http://www.wolfenet.com/~jeremiah Received on Wed Oct 07 1998 - 10:18:20 CDT