lru error: link
- From: j_chang@xxxxxxxxxxx
- To: oracle-l@xxxxxxxxxxxxx
- Date: Thu, 14 Oct 2010 03:24:15 -0700
Hello,
Can anyone shed any light on the following message that appeared in
Oracle trace files:
LRU ERROR: LINK=(some number)
This was accompanied by ORA-600 [17126], ORA-7445 [Address not mapped to
object] with repeated core dumps several times a minute until the
database was restarted. On 10.2.0.4 Solaris 64 bit. Oracle Support
recommends changing _kghdsidx_count to 1 or _kgl_latch_count to <=31,
but give no indication of how this connects with the error message, let
alone what could have caused the "LRU ERROR." What does this error
mean? The top wait event during the period of dumping core was for
library cache latch, so was it some error with the LRU list for buffers
in the library cache (as opposed to the buffer cache)? What type of
error could an LRU list encounter? The 2nd highest wait event was for
db file sequential read. Could this be an issue with exceptionally high
i/o? (though there are earlier occasions when db file sequential (and
scattered) read had even higher waits) Any ideas welcome; more details
can beprovided.
Thanks,
--
http://www.fastmail.fm - A no graphics, no pop-ups email service
--
//www.freelists.org/webpage/oracle-l
Other related posts:
