Connection to backup database lost - MYSQL$BKUPEXEC ? 2

[TheTraveler]

Hi all,
Thank you for helping if possible.

I have a fresh copy of Backup Exec 9 installed on my Windows 2000 server. It's a single server backup, single tape, local backup. Whenever I set up a job to run overnight, I come in the morning only to see the job is still sitting in "Ready" status. When I open up the Management interface, I get the message "The connection to the Backup Database has been lost." When I check services, the MYSQL$BKUPEXEC service is not running. I can never get back into the management interface again until I restart all the backup services.

I've set the service to restart itself, but it never does. I've tried uninstalling everything including saved jobs, and reinstalling, didn't help. I've tried it with 9.0 and 9.1 to no avail. Does anyone have a clue on this one?

I used to have 8.6 running on the server and it ran ok. It had other issues, but not this. This is a full version I purchased, so I completely uninstalled 8.6 and installed 9.

Thanks for any pointers.

 

[SQLBill]

First things first - Error messages

Are there any errors in the Windows Event Viewer logs?

Once you get the management window working, right click on the failed job, select properties, select Job History tab. Any additional errors there?

-SQLBill

 

[TheTraveler]

Thanks for taking a look at my issue, SQLBill.

The events are as follows.

1. I scheduled a job to run at 8:00.

2. At 8:00 I get the following Windows Events in Application for BKUPEXEC:

EVENT
Adamm Database Error: Connection Lost!
Error = E_PVL_DB_LOST_CONNECTION
Server = "DATA"
Active Node = ""
Instance = "BkupExec"
Database = "BEDB"

EVENT
database connection failed
For more information, click the following link:

http://eventlookup.veritas.com/eventlookup/EventLookup.jhtml

EVENT
Access to catalog index (Catalog index database) failed.
Reason: Null connection object pointer cat_RecordSet::cat_RecordSet()
d:\be\tybee\4367r\becat\server\dll\..\catalogindex.cpp(1689)

EVENT
Access to catalog index (Catalog index database) failed.
Reason: Null connection object pointer cat_RecordSet::cat_RecordSet()
d:\be\tybee\4367r\becat\server\dll\..\catmediaiterator.cpp(218)

3. I check the services, and the MSSQL$BKUPEXEC service is no longer running.

4. If I restart the MSSQL service, I still can't get into the management interface, I have to restart all the backup exec services.

5. Once I restart all services and reenter the interface, I check the job status, and there is sits, waiting in a "ready" status.

6. I don't get any BackupExec errors because it doesn't think that the job has failed. It's still sitting there waiting to begin.

7. In this particular test instance, I tried to manually run the job "Run Now", and the MSSQL service stopped again, and I received the same "Connection to the backup database has been lost" message. I then restarted the services again, and it finally started running. It seems random as to when the job will actually run.

Any help is greatly appreciated. Thanks!

 

[SQLBill]

It looks to me like Backup Exec is having problems with the catalog. If it can't access the catalog, it can't record the data needed for restores and that will cause it to fail.

Have you tried using the Installation disk and choosing REPAIR?

A file/folder with the catalog information might have been moved/deleted.

BTW-on my BE 9, I show the service as set to Automatic and LocalSystem Login.

-SQLBill

 

[TheTraveler]

I am showing MYSQL$BKUPEXEC as running Automatic and LocalSystem login, all the other backup exec services are running Automatic and using the Admin login.

I ran the Repair option from the installation disk, and get the same issues.

 

[SQLBill]

I did a search on my BE9 server for the two files your EVENT errors are having problems finding:

d:\be\tybee\4367r\becat\server\dll\..\catalogindex.cpp(1689)
and
d:\be\tybee\4367r\becat\server\dll\..\catmediaiterator.cpp(218)

But I can't find any files with those names on my BE9 server. HMMMMMM interesting.

I did find a BeCatSrv.dll file (several in fact), but no becat folder (nor server, nor dll folders).

-SQLBill

 

[justin2000]

Try creating a new db -

http://seer.support.veritas.com/docs/254014.htm

if that fails, which it might, you may have mismatched dll's in MDAC.

you can try rolling back the mdac - i think it's dasetup /u (search for dasetup on the computer), change to this directory in DOS and run the command.
reboot the server.
then install MDAC again, its on the cd or download - winnt\install\mdac
reboot.

if you get the same error
check the data path regkey in HKLM\Software\Veritas\backup Exec\engine\misc - make sure it is point to \program files\veritas\backup exec\nt\data directory

 

[TheTraveler]

Bill, I just noticed that those 2 catalog files its searching for, its looking on D: thats my CD drive! huh? Any thoughts on that?

Justin, thank you for the tips. I'm trying them now. I checked the registry path and it is pointing to the correct \program files\veritas\backup exec\nt\data directory.

I also tried to run the "Recovery" from the support document link you posted, and heres what I got:

Starting database recovery.
Stopping Services
Stopping services for serverATA.
ServerATA, Service:ECM Service stopped successfully or was not running.
ServerATA, Service:BackupExecAgentBrowser stopped successfully or was not running.
ServerATA, Service:BackupExecJobEngine stopped successfully or was not running.
ServerATA, Service:BackupExecRPCService stopped successfully or was not running.
ServerATA, Service:BackupExecDeviceMediaService stopped successfully or was not running.
ServerATA, Service:BackupExecNamingService stopped successfully or was not running.
ServerATA, Service:BackupExecAgentAccelerator stopped successfully or was not running.
Stopping services for server DATA completed.
Performing database recovery.
Error: Unable to attach database files.
Database recovery for server DATA ended with errors.

It doesn't restart the services or anything and when I manually restart services and go back into backup exec, its all the same.


I probably wont get a chance to run the rollback of MDAC until this evening because this server is actively used during the day (I snuck in a reboot to try a "repair" option from the CD early though but only because everyone was in a meeting).


Thanks guys for the suggestions, I feel like I'm getting a lot closer than I have been in the last few weeks of fighting this issue.

 

[TheTraveler]

Rebuilding the MDAC seems to have worked!!!! I was able to run a manually started test backup and a scheduled test backup this morning. I won't feel 100% sure until this evening when my regular scheduled backup tries to commence, but it looks good!

Justin2000, I can't thank you enough! Thanks to you too for helping me along SQLBill. I've learned some valuable troubleshooting tips along the way. You guys rock.