Possible BUG in Oracle CRS stack for linux
Hi!
This is my first post here, so sorry for any "errors" ... :-)
Now, on to the problem. I have two clasters (Oracle RAC) - one on pair of Dell servers, and one on pair of HP servers.
I did install Oracle RAC (Grid and DB) on both of them successfully, BUT i did notice one difference.
It's 11.2.0.2.0 version of CRS and database. OS is RHEL 5.5, 64bit.
CRS stack on Dell servers produced significantly more disk I/O on installed disk than one on HP. After investigating, i did manage to trace it down to $CRS_HOME/crf/db/<host name>/ directory. On Dell servers, this directory had big "dbt" files, on both nodes, and current update time. On the other hand, on HP servers, only on installation server there were files (old!), while on other node (2 node cluster) - only one file: <hostname>.ldb (also old).
This is my first post here, so sorry for any "errors" ... :-)
Now, on to the problem. I have two clasters (Oracle RAC) - one on pair of Dell servers, and one on pair of HP servers.
I did install Oracle RAC (Grid and DB) on both of them successfully, BUT i did notice one difference.
It's 11.2.0.2.0 version of CRS and database. OS is RHEL 5.5, 64bit.
CRS stack on Dell servers produced significantly more disk I/O on installed disk than one on HP. After investigating, i did manage to trace it down to $CRS_HOME/crf/db/<host name>/ directory. On Dell servers, this directory had big "dbt" files, on both nodes, and current update time. On the other hand, on HP servers, only on installation server there were files (old!), while on other node (2 node cluster) - only one file: <hostname>.ldb (also old).
0