Hi,
I've lost my network connection temporarily, so can't access Metalink.
Real quickly on ORA-1555s...
1. For consistent reads, the buffer cache will be searched prior to the
rollback segments. I don't think it likely, but wouldn't reducing
db_block_buffers then make it less probable, if only slightly, that the
required info to reconstruct a read-consistent view will not be available?
And yes, it definitely will still have to look at the rollback segments and
the slots for the rollback segments.
2. For delayed block cleanout problems potentially leading to ORA-1555s, I
suspect that a FTS could reduce the likelyhood of problems with the data.
Short of constructing queries to go after each of the available indices,
could we use an analyze of the indexes to avoid the issue on the index side?
Any gotchas?
3. In the past a FTS added data to the MRU end of the data buffer. I have
someone telling me that is no longer true with 8.0, as in it doesn't go into
the databuffer at all. Anyone?
Received on Tue Sep 12 2000 - 00:16:35 CDT
