These are the following Errors/Warning messages which we will get in 10.2.0.1 release Instances alert log files.  Find the following work arounds/solutions, what I have successfully applied.

 Problem 1:

 Memory Notification: Library Cache Object loaded into SGA
Heap size 6237K exceeds notification threshold (2048K)
 

These are warning messages only which should not cause any problems. These warning messages appear due to new event messaging mechanism and memory manager in 10g Release 2. In 10g R2 we have a new undocumented parameter that sets the KGL heap size warning threshold. Warnings are written if heap size exceeds this threshold. 

Set    _kgl_large_heap_warning_threshold to a reasonable high value to prevent these warning messages: Default Threshold in 10.2.0.1 is 2 MB only. But Default Threshold value in 10.2.0.2 is 50 MB which is reasonable and recommended value.  

Solution: 

SQL> alter system set “_kgl_large_heap_warning_threshold”= 52428800 scope=spfile;

SQL> shutdown immediate; 

SQL> startup;   

Problem 2: 

Trace file created when RMAN issues the controlfile autobackup. 

Following messages will be appeared in alert log file 

Errors in file /oracle/product/10.2.0.1/admin/udump/prod_ora_18627.trc:
Control autobackup written to DISK device
handle ‘/backup/flash_recovery_area/PROD/autobackup/2008_01_04/o1_mf_s_648508685_3wv4m226_.bkp’ 

 These messages do not have any impact. Only problem is when ever RMAN issues the control file autobackup command, it will create one trace file with the above errors.  

Work around  

(1)   This is a bug fixed in 10.2.0.2 or 10.2.0.3       

  Or       

(2) Write an OS cron job to delete the trace files 

 References:

330239.1. Memory Notification: Library Cache object loaded into SGA 

 Note: If you are having above problems with your instances, Please login as SR to Oracle Support and then implement the solution under their guidance.