Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: drop user user_name cascade - question
> Depending on the block size of the database (parameter db_block_size,
> usually 8k)
Yes, it's 8k
> your setting of db_block_buffers looks way too high (some
> 560 M) and you also have some 200M shared pool. You must be running
> out of virtual memory, and this suspicision is underpinned by the core
> dumps you are getting. The guideline of Oracle is to limit the cache
> to one third of *physical* RAM, your case some 600M
> You may also suffer from an insufficient log_buffer parameter, an
> insufficient sort_area_size parameter (the default is 64k, which is
> just, ahem, ridiculous), and incorrectly sized online redolog files
> (in this particular case you need to check for 'checkpoint not
> complete' and 'cannot allocate log, archival required' messages in
> your alert.
It this case, could you pleaes recommend me something, so that I can try:
db_block_buffers = ?
shared_pool_size = ?
sort_area_size = ?
incorrectly sized online redolog files = set what, where?
> Are the affected objects in a locally managed tablespace or a
> dictionary managed tablespace? In the latter case, you will be getting
> so-called space transactions (which involves updating sys.uet$ and
> sys.fet$) using the ST lock, of which there can be only one.
How do I check that?
Thanks in advance!
Tom Received on Sun Oct 13 2002 - 10:00:46 CDT