ORA-07445 Exception Encountered - Core Dump Analysis Guide

ORA-07445: Exception Encountered

Error Overview

Error Text: ORA-07445: exception encountered: core dump [string] [string] [string] [string] [string] [string]

The ORA-07445 error indicates an Oracle process has encountered an exception causing a core dump. This is a serious error typically caused by software bugs, memory corruption, or OS-level issues. Like ORA-00600, this usually requires Oracle Support assistance.

Understanding ORA-07445

Error Structure

ORA-07445: exception encountered: core dump [kglic0()+1100] [SIGSEGV] [ADDR:0x0] [PC:0x1234567] [Address not mapped to object] []
                                            ^^^^^^^^^^^^^^^^  ^^^^^^^^
                                            Function+offset   Signal type

Common Signal Types

Signal	Description	Common Cause
SIGSEGV	Segmentation violation	Invalid memory access
SIGBUS	Bus error	Memory alignment issues
SIGFPE	Floating point exception	Division by zero
SIGILL	Illegal instruction	Corrupted code
SIGABRT	Abort signal	Process self-termination

Initial Response

1. Assess Impact

-- Check if database is still running
SELECT instance_name, status, database_status
FROM v$instance;

-- Check for other affected sessions
SELECT COUNT(*) as active_sessions
FROM v$session
WHERE status = 'ACTIVE' AND type = 'USER';

-- Recent errors
SELECT originating_timestamp, message_text
FROM x$dbgalertext
WHERE message_text LIKE '%ORA-07445%'
   OR message_text LIKE '%ORA-00600%'
ORDER BY originating_timestamp DESC
FETCH FIRST 20 ROWS ONLY;

2. Locate Core Dump and Trace Files

# Find core dump location
cd $ORACLE_BASE/diag/rdbms/$ORACLE_SID/$ORACLE_SID/cdump
ls -la core*

# Alternative locations
find $ORACLE_BASE -name "core*" -mtime -1 2>/dev/null

# Find associated trace file
cd $ORACLE_BASE/diag/rdbms/$ORACLE_SID/$ORACLE_SID/trace
ls -lt | grep -E "(ora_|j00)" | head -20

# Check incident
cd $ORACLE_BASE/diag/rdbms/$ORACLE_SID/$ORACLE_SID/incident
ls -la

3. Gather Diagnostic Information

# Using ADRCI
adrci

ADRCI> set home diag/rdbms/mydb/mydb
ADRCI> show incident -mode detail -p "error_text like '%07445%'"

# Create incident package
ADRCI> ips create package incident <incident_id>
ADRCI> ips add file <core_dump_path> package 1
ADRCI> ips generate package 1 in /tmp/ora7445_package

# Manual collection
tar -czf ora7445_diag.tar.gz \
  $ORACLE_BASE/diag/rdbms/$ORACLE_SID/$ORACLE_SID/trace/*ora*.trc \
  $ORACLE_BASE/diag/rdbms/$ORACLE_SID/$ORACLE_SID/cdump/core* \
  $ORACLE_BASE/diag/rdbms/$ORACLE_SID/$ORACLE_SID/alert/alert*.log

Common Scenarios and Solutions

Scenario 1: Memory Corruption

-- Check memory parameters
SHOW PARAMETER memory_target
SHOW PARAMETER sga_target
SHOW PARAMETER pga_aggregate_target

-- Flush memory structures
ALTER SYSTEM FLUSH SHARED_POOL;
ALTER SYSTEM FLUSH BUFFER_CACHE;

-- If recurring, reduce memory to avoid bad memory region
ALTER SYSTEM SET sga_target = 8G SCOPE=SPFILE;
ALTER SYSTEM SET pga_aggregate_target = 2G SCOPE=SPFILE;
-- Restart required
SHUTDOWN IMMEDIATE;
STARTUP;

Scenario 2: OS Resource Issues

# Check system resources
free -m
df -h
ulimit -a

# Check for OS errors
dmesg | tail -50
grep -i error /var/log/messages | tail -20

# Verify Oracle user limits (/etc/security/limits.conf)
oracle soft nproc 2047
oracle hard nproc 16384
oracle soft nofile 1024
oracle hard nofile 65536
oracle soft stack 10240
oracle hard stack 32768

Scenario 3: Known Bug Patterns

-- Common bug indicators
-- Function names often indicate the area:
-- kgl* - Library cache
-- kks* - SQL Execution
-- ktb* - Table access
-- ktu* - Undo management
-- kcb* - Buffer cache

-- Check for known patches
SELECT patch_id, action, status, description
FROM sys.dba_registry_sqlpatch
WHERE description LIKE '%07445%'
ORDER BY action_time DESC;

Workarounds and Temporary Fixes

1. Disable Problematic Features

-- Common feature workarounds
ALTER SYSTEM SET "_cursor_features_enabled"=2 SCOPE=BOTH;
ALTER SYSTEM SET "_optimizer_adaptive_plans"=FALSE SCOPE=BOTH;
ALTER SYSTEM SET "_px_adaptive_dist_method"='OFF' SCOPE=BOTH;

-- Disable specific optimizer features
ALTER SYSTEM SET "_optimizer_null_aware_antijoin"=FALSE SCOPE=BOTH;
ALTER SYSTEM SET "_optimizer_ansi_rearchitecture"=FALSE SCOPE=BOTH;

2. Session-Level Workarounds

-- For specific sessions experiencing issues
ALTER SESSION SET optimizer_mode = RULE;
ALTER SESSION SET "_complex_view_merging"=FALSE;
ALTER SESSION SET workarea_size_policy = MANUAL;
ALTER SESSION SET sort_area_size = 524288000;

3. SQL-Level Workarounds

-- Use hints to avoid problematic code paths
SELECT /*+ NO_MERGE USE_NL(t1 t2) */
  t1.col1, t2.col2
FROM table1 t1, table2 t2
WHERE t1.id = t2.id;

-- Rewrite problematic queries
-- Instead of complex MERGE
MERGE INTO target t
USING source s ON (t.id = s.id)
WHEN MATCHED THEN UPDATE...
WHEN NOT MATCHED THEN INSERT...;

-- Use separate UPDATE and INSERT
UPDATE target t
SET ...
WHERE EXISTS (SELECT 1 FROM source s WHERE s.id = t.id);

INSERT INTO target
SELECT * FROM source s
WHERE NOT EXISTS (SELECT 1 FROM target t WHERE t.id = s.id);

Analysis Techniques

Stack Trace Analysis

# Extract function call stack
grep -A 20 "Call Stack" <trace_file>

# Common problematic call stacks
# Buffer cache issues:
kcbgtcr -> kcbgcur -> ktbgcur -> ktrget

# Library cache issues:
kglLock -> kglget -> kksfbc -> kkspsc0

# Memory issues:
kghalf -> kghalp -> kghfnd -> kghalo

Using GDB for Core Analysis

# Basic core dump analysis
gdb $ORACLE_HOME/bin/oracle core.<pid>

(gdb) where
(gdb) info registers
(gdb) x/10i $pc-20
(gdb) quit

# Extract readable information
strings core.<pid> | grep -E "(SELECT|INSERT|UPDATE|DELETE)" | head -20

Preventive Measures

System Health Monitoring

-- Create monitoring procedure
CREATE OR REPLACE PROCEDURE monitor_system_health AS
  v_count NUMBER;
BEGIN
  -- Check for recent ORA-07445
  SELECT COUNT(*) INTO v_count
  FROM x$dbgalertext
  WHERE message_text LIKE '%ORA-07445%'
    AND originating_timestamp > SYSTIMESTAMP - INTERVAL '1' HOUR;

  IF v_count > 0 THEN
    -- Send alert
    DBMS_OUTPUT.PUT_LINE('CRITICAL: ORA-07445 detected in last hour');
    -- Can integrate with email/monitoring system
  END IF;

  -- Check core dump directory
  -- Use UTL_FILE to check for new core files
END;
/

-- Schedule monitoring
BEGIN
  DBMS_SCHEDULER.CREATE_JOB(
    job_name => 'MONITOR_SYSTEM_HEALTH',
    job_type => 'STORED_PROCEDURE',
    job_action => 'monitor_system_health',
    repeat_interval => 'FREQ=HOURLY',
    enabled => TRUE
  );
END;
/

Patch Management

# Regular patch checks
$ORACLE_HOME/OPatch/opatch lsinventory -detail > patch_inventory.txt

# Check for recommended patches
# Search MOS for:
# - Database version + "recommended patches"
# - "ORA-07445" + first argument function
# - Platform-specific bundle patches

# Apply patches proactively
cd $ORACLE_HOME
$ORACLE_HOME/OPatch/opatch prereq CheckConflictAgainstOHWithDetail -ph <patch_location>
$ORACLE_HOME/OPatch/opatch apply <patch_location>

Oracle Support Interaction

Information to Provide

1. Complete error message from alert log
2. Stack trace from trace file
3. Core dump file (if under 2GB)
4. System information:

   uname -a
   cat /etc/os-release
   free -m
   ulimit -a

5. Recent changes (patches, upgrades, application changes)

SR Template for ORA-07445

Problem: ORA-07445 [function_name]

Environment:
- Database Version: 19.15.0.0
- OS: Oracle Linux 7.9
- Platform: x86_64

Error Frequency: [Once/Intermittent/Frequent]
First Occurrence: [Date/Time]
Business Impact: [Critical/High/Medium]

Error Details:
ORA-07445: exception encountered: core dump [full_error_text]

Associated SQL: [If available from trace]

Troubleshooting Done:
1. [List steps taken]
2. [Workarounds attempted]

Diagnostic Files Uploaded:
- Incident package: [filename]
- Core dump: [filename]
- Alert log: [filename]

Recovery Procedures

Instance Recovery

-- If instance crashed
STARTUP MOUNT;
ALTER DATABASE RECOVER;
ALTER DATABASE OPEN;

-- Check for corruption
RMAN> VALIDATE DATABASE;

-- Review alert log for additional errors
SELECT originating_timestamp, message_text
FROM x$dbgalertext
WHERE originating_timestamp > SYSTIMESTAMP - INTERVAL '1' HOUR
ORDER BY originating_timestamp;

Session Cleanup

-- Find and clean up affected sessions
SELECT s.sid, s.serial#, s.username, s.program, s.status,
       p.spid, p.pid
FROM v$session s, v$process p
WHERE s.paddr = p.addr
  AND s.status = 'KILLED';

-- Force cleanup if needed
ALTER SYSTEM KILL SESSION 'sid,serial#' IMMEDIATE;

-- OS level cleanup
-- kill -9 <spid>

Related Errors

• ORA-00600 - Internal error code
• ORA-03113 - End-of-file on communication channel
• ORA-00603 - ORACLE server session terminated
• ORA-04030 - Out of process memory

Quick Reference

Emergency Response Checklist

1. ✓ Check database status
2. ✓ Locate core dump and traces
3. ✓ Create incident package
4. ✓ Check for known bugs
5. ✓ Apply workarounds if available
6. ✓ Open Oracle SR with Severity 1 if production
7. ✓ Monitor for recurrence
8. ✓ Plan patching strategy

Common Functions and Areas

• kgl* - Library Cache
• kks* - SQL Compilation
• kxs* - SQL Execution
• kcb* - Buffer Cache
• ktb* - Table Access
• ktu* - Undo/Transaction
• kge* - Error Handling
• kgh* - Memory Management