One-time processing of the adventures of ORA-07445)

The ORA-07445 is typically the result of an Oracle call to an operating system resource error [@ [email protected]

There was no warning beforehand. At about five o'clock P.M., database instances on 17 Oracle databases of a key application went down one after another. Check alert_log and find that this file contains a large number of 0ra-07445 error codes: ORA-07445: exception encountered: core dump [] [] [] [] [] [] [] also do not look for root cause, hurry to restart the database, however, even worse, the database instance fails to be started. The database instance can be started successfully only once in a row after multiple startups, and the database is quickly down, And the listener often fails to be started. First, the server is infected with viruses. The application environment is Oracle 10.2.0.1 and Windows 2003 Server. Because the ORA-07445 usually occurs when Oracle calls an operating system's resource error. After checking the Oracle parameters, I was surprised to find that the database was running in the sharing mode. I quickly changed all of them to the dedicated mode. The related statement is as follows: Alter system set dispatchers = '; alter system set shared_servers = 0; restart the database again. The database can be started, but the instance will occasionally go down, but it will be restarted immediately, in this way, restart the database in a few minutes. Finally, I got off work, and the phone around me was still quiet. Now I can start to calm down and find root cause. From the preceding symptom analysis, it is obvious that the OS is faulty and the Oracle database fails to call Windows 2003 Server resources, Because the dedicated mode reduces the interaction between Oracle and OSSo the downtime is reduced. Go to Metalink to find a similar problem and find two documents: Doc ID 422471.1 and Doc ID: 405904.1. After analysis, the following two changes were implemented: Change 1: to reduce the number of accesses to the database and OS, block the authentication of the OS login Database: In sqlnet. in ora, the following statement is commented: # sqlnet. authentication_services = (ETS) Change 2: To speed up the database response to the login session, modify the following listener parameter sqlnet. add the following statement sqlnet. inbound_connect_timeout = 0 --- the default value is 60 seconds in listener. add inbound_connect_timeout_listener = 0 in ora --- the default value is 60 seconds.

It has the following features:

• The svchost.exeerror is reported for 15 machines at the same time point in the first day. The svchost.exe error is also generated by multiple machines at the same time.
• In conjunction with oracle's alert_loganalysis, The svchost.exe error soon, the database encountered a ora-07445 error and then went down.
• Error Module kernel32.dll, error address 0x0010568f.

You can find instructions and solutions for this vulnerability on the following two web pages of windows. Ipvw32.downadup. B-type worm, must be patched with kb958644. After the patch is punched in, the problem is solved. Summary: a computer on the Intranet is infected with viruses, which constantly attack the svchost process of Windows systems in the same network segment, resulting in Oracle downtime. It can be solved by adjusting Oracle parameters and patching windows. Oracle Metalink Doc ID: 422471.1 and Doc ID: 405904.1 windows: http://www.microsoft.com/china/technet/security/bulletin/MS08-067.mspxhttp://support.microsoft.com/kb/958644/zh-cn http://blog.itpub.net/7695891/viewspace-1040157/

One-time processing of the adventures of ORA-07445)