Alert fired twice

valoche · Post by **valoche** » 27 Aug 2007 11:03

Hi,

I have a recurring problem with some jobs. They end in a not ok state, so an alert is triggered.

Problem is sometimes, a second alert is triggered several days later, for no reason.

I have no ideas why this happens, before calling the support I'd like to have your opinion.

Thanks.

Here is an example :

ALERT 1 :
Fri Aug 17 21:35:46 BST 2007
PARAMETRES: call_type: I alert_id: 40756 data_center: XXXXXX memname: QU-EUFR-BKP-HR-ORA-J017-STO006 order_id: 045mj severity: V status: Not_Noticed send_time: 20070817213546 last_user: emuser last_time: 20070314112206 message: Ended not OK owner: oracle group: QU-EUFR-BKP-PPS-D01 application: QU-EUFR-BKP job_name: QU-EUFR-BKP-HR-ORA-J017-STO006 node_id: XXXXXXX

ALERT 2 :
Thu Aug 23 06:03:13 BST 2007
PARAMETRES: call_type: I alert_id: 42422 data_center: XXXXXX memname: QU-EUFR-BKP-HR-ORA-J017-STO006 order_id: 045mj severity: V status: Not_Noticed send_time: 20070823060313 last_user: emuser last_time: 20070314112206 message: Ended not OK owner: oracle group:
QU-EUFR-BKP-PPS-D01 application: QU-EUFR-BKP job_name: QU-EUFR-BKP-HR-ORA-J017-STO006
node_id: XXXXXXX

JOB DEFINITION :
$ ctmpsm -LISTDETAILS 045mj
Job full details for order number:000045mj

Jobname: QU-EUFR-BKP-HR-ORA-J017-STO006
Mem Name: QU-EUFR-BKP-HR-ORA-J017-STO006
Group: QU-EUFR-BKP-PPS-D01
Application: QU-EUFR-BKP
Owner: oracle
NodeGroup:XXXXXXX
Task Type: C
Cyclic: N
Priority: AA
Critical: N
Max Rerun: 0
Interval: 00000M

Time From: 2120
Until:
Time Zone:
Max Wait: 99
Mem Lib :
Over Lib:
Sysout Option:
Sysout Parm:
Command Line:/XXXXXXXXX/exploit/oracle/pl/hotbackup.pl HR89PPD2 begin

philmalmaison · Post by **philmalmaison** » 27 Aug 2007 2:14

it look like a bug, but if you don't want old alerts, you can daily remove them.
You can do it by the following command :
erase_alerts -U <user_ecs> -P <password_ecs> -D <aaaammjj> -F
where -D is the date from wich you keep alerts.

ofen this date is replace by an autoedit parm (giving $1 in command line) :
erase_alerts -U <user_ecs> -P <password_ecs> -D $1 -F

and

%%PARM1=%%$CALCDATE %%$DATE -n
where n is the number of days to keep.

for your prblem don't hesitate to make a BMC call

valoche · Post by **valoche** » 27 Aug 2007 3:52

Ok great.
I looked closer, we indeed purge the alerts, keeping 5 days. I lowered the threshold to 2 days, because we don't really need the alerts history.

Cheers.

mauriziog · Post by **mauriziog** » 28 Aug 2007 11:25

Is also possible to disable the alerts send if the job is finished notok, is done in the system parameters. But in this manner you havent the alert if the job finished notok, so is the solution if the alerts are not used or used for other purposes.