Link to home
Start Free TrialLog in
Avatar of ThomasBell
ThomasBell

asked on

Help with Oracle Enterprise Manager backup...getting "Could not fork or create job process." error.

I am attempting to create a backup job and it keeps failing.

I have tried rebooting the machine, etc. and nothing helps.

The only error I get in the "View Output" window is "Could not fork or create job process."

I am running Personal Oracle9i Release 9.2.0.1.0 - Production.

Thanks.
Avatar of Danielzt
Danielzt

if you are on unix, refer to this:

fix:
Reset/configure the kernel to increase the maximum number of processes
available to each user. This can be done on AIX base systems by doing the
following:

1. Run smit
2. Select Change/Show characteristics of Operating System.
3  Select maximum number of processes allowed per user.
Avatar of ThomasBell

ASKER

I should have explained that I am using Windows XP Pro...sorry for omitting that.
Are there any additional messages in dbsnmp*.log?

Do you have other jobs that run successfully?

Can you connect using dbsnmp?  sqlplus dbsnmp/dbsnmp@sid
This is the only job type that I have attempted to run so the answer to your question about other jobs being successful is no.

I have attempted to run a backup at least 10 times, all failing with this error.

Yes, I can connect using "sqlplus dbsnmp/dbsnmp@develop" with no errors.

You asked, "Are there any additional messages in dbsnmp*.log?"

Here is the output today from the DBSNMP.LOG file:

-- xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

DBSNMP for 32-bit Windows: Version 9.2.0.1.0 - Production on 08-APR-2004 11:57:18

Copyright (c) 2002 Oracle Corporation.  All rights reserved.

System parameter file is C:\Oracle\Ora81\network\admin\snmp_ro.ora
Log messages written to C:\Oracle\Ora81\network\log\dbsnmp.log
Trace information written to C:\Oracle\Ora81\network\trace\dbsnmp_1452.trc
Trace level is currently 0

NMS-00001: Warning: dbsnmp unable to connect to SNMP master agent
08-APR-2004 14:10:39: NMS-00908: Shutting down normally

DBSNMP for 32-bit Windows: Version 9.2.0.1.0 - Production on 08-APR-2004 14:16:45

Copyright (c) 2002 Oracle Corporation.  All rights reserved.

System parameter file is C:\Oracle\Ora81\network\admin\snmp_ro.ora
Log messages written to C:\Oracle\Ora81\network\log\dbsnmp.log
Trace information written to C:\Oracle\Ora81\network\trace\dbsnmp_4992.trc
Trace level is currently 0

NMS-00001: Warning: dbsnmp unable to connect to SNMP master agent

DBSNMP for 32-bit Windows: Version 9.2.0.1.0 - Production on 08-APR-2004 14:30:55

Copyright (c) 2002 Oracle Corporation.  All rights reserved.

System parameter file is C:\Oracle\Ora81\network\admin\snmp_ro.ora
Log messages written to C:\Oracle\Ora81\network\log\dbsnmp.log
Trace information written to C:\Oracle\Ora81\network\trace\dbsnmp_1768.trc
Trace level is currently 0

NMS-00001: Warning: dbsnmp unable to connect to SNMP master agent

-- xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Avatar of schwertner
What kind of Backups are you trying to do?
If you do not use Archivelog mod the only kind of backup is FULL COLD Backup which also means that you have to shutdown the Oracle Instance.
Be aware that Personel Oracle (even Standart Edition of Oracle) comes with restricted functionality. So if you would like to do complex DBA tasks you have to use Enterprise Edition of Oracle.
DESTROY
Default: N

Specifies whether or not the existing data files making up the database should be reused. That is, specifying DESTROY=Y causes Import to include the REUSE option in the datafile clause of the CREATE TABLESPACE command, which causes Import to reuse the original database's data files after deleting their contents.

Note that the export file contains the data file names used in each tablespace. If you specify DESTROY=Y and attempt to create a second database on the same machine (for testing or other purposes), the Import utility will overwrite the first database's data files when it creates the tablespace. In this situation you should use the default, DESTROY=N, so that an error occurs if the data files already exist when the tablespace is created. Also, when you need to import into the original database, you will need to specify IGNORE=Y to add to the existing data files without replacing them.

Caution: If data files are stored on a raw device, DESTROY=N does not prevent files from being overwritten.

Interesting.... I`m testing this for you. ;)
opps.. sorry.
GDE, is your post intended for this question?  If so, I don't understand what you are telling me...
schwertner,

The database is in NOARCHIVELOG mode but according to the Backup Wizard, if the database is open at the time the job runs, it will be shutdown for the backup and "brought back up" once the backup is complete...
schwertner,

I changed the mode to Archive mode and the backup still fails with the exact same error.  Hope this info helps.
my post not for this question,  i`m miss.
sorry.

Best regards.
Dmitry.
may be increase job_queue_processes in init file?

Best regards.
Dmitry
GDE, I increased the "job_queue_processes" from 10 to 50 and it still doesn't work.

While I have been waiting for a solution I have also tried doing an import (IMP) using the Enterprise Manager console and it gives the same error.
@ThomasBell :

1. "Could not fork or create job process." : Seems like more of OS error ...check your Event Viewer to get more information.

2. Do you have enough resources on machine where your OEM managment server is running ?. Check you memory usage ..others resource usgae during job creation..

3. If possible post "Event Viewer " content assoicated with OEM applicaitons if any... for furthur comments....

-Raj
ASKER CERTIFIED SOLUTION
Avatar of GDE
GDE

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
GDE, I can already do an IMP and EXP from a command prompt -- it works just fine, and could probably work around the problem of scheduling by using the AT command as you suggest.

I really want to use OEM for this though, as it should work.

Thanks.
We are having the same problem.  It doesn't have anything to do with backup it has to do with connectivity between OEM and the target server.  I can run jobs on several boxes but we have a brand new HP RAC system built by HP that is throwing this error.   We were able to discover the node and the db.  The preferred credentials are set and the node user has log on as batch and log on as service.  Has any progress been made on this?  There doesn't seem to be any info on metalink that is helpful.  Our logfile looks like the one above.  The error shown is common and not the problem.
And, we are on Windows 2003 Advanced Server and Oracle 9.0.2.4.
SOLUTION
Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
I am having the same problem and I have tried all the solutions provided above but still the stuck.

Please guide.
Regards
Asim