Version

All

Date

20 Aug 2010

Summary

A Daughter Engine’s Supervising Engine service may shut itself down, or show the following signs in the Supervising Engine logs:

04 Aug 2010 02:35:20.673 SVR1 admin Successfully downloaded control information 

                                    'c:\argent\ArgentManagementConsole\CB_DATA.BKU' from remote server. 

04 Aug 2010 02:35:20.673 SVR1 admin DOWNLOADED 0.013828 MB Shared Database 

04 Aug 2010 02:35:20.688 SVR1 admin Could not to restore the file 

                                    c:\argent\ArgentManagementConsole\CB_DATA.BKU 

                                    (The filename, directory name, or volume label syntax is incorrect.)(123)

Sometimes, the CB_DATA.BKU may appear to be restored successfully, then a minute later:

04 Aug 2010 02:36:20.673 SVR1 admin FORCED to initialize relator information

04 Aug 2010 02:37:22.673 SVR1 admin The exit event of parent service is opened successfully.

These chunks of logs may keep repeating itself, with the Daughter Supervising Engine intermittently stopping on its own.

Technical Background

In a Mother/Daughter architecture, the Daughter downloads a file called CB_DATA.BKU that contains information on the “shared database”.

The “shared database” consists of all shared definitions — Monitoring Groups, Alerts, Calendars, etc .

If the local copy of the shared database somehow becomes corrupted, it can wreak havoc and cause issues with the Daughter Supervising Engine.

Reasons for corruption may include antivirus, disk issues, or network issues.

Downloading the shared database actually comes in two modes:

1. Downloading only the incremental changes in the shared database at each poll (default)

2. Downloading the complete shared database at each poll

If the shared database becomes corrupted, downloading only the incremental changes does nothing to an already corrupted system.

The downloaded data is expanded into the Daughter Engine’s product DBF and DATA folders.

Resolution

If shared database corruption is suspected, then we need clear the corrupted database files, and change the download mode to download the complete shared database at each poll.

Do the following steps:

1. Stop all services on the Daughter Engine

2. Move out all the files under the DATA folder into a backup folder

3. Move out all the files under the DBF folder into a backup folder — EXCEPT the ENGINE.* files.

(The ENGINE.* files are required to identify the Daughter Engine to the Mother)

4. Change the registry key on the Main Engine only:

HKLM\Software\Argent\ArgentManagementConsole\ArgentAlertConsole\CDATA_1

From 0 to 1

(0 means incremental changes, 1 means always download the complete shared database)

5. Restart the Argent Alert Console service on the Main Engine only, and verify that the identified log excerpts no longer exist.