Posted by 
The section of T24 OPERATIONS is as much important as the rest of the T24 Modules. During the COB/EOD -which is the most critical part of OPERATIONS- many activities are taking place. The operator has to be focused and to monitoring the output of the COMO file as is vital to identify on time a possible locking of the a JOB LIST so as to report it to CSHD of Temenos.
Detailed Content
COB Monitoring
For the monitoring of the COB there are some important tools.
-
- T24 desktop command ENQ AGENT.STATUS which shows the number of the agents running the COB and the relevant job that each agent runs.
- COB.MONITOR. This command given via mainline routine could provide a general overview of the % progress of the COB.
This display a list of comos like tSA_10_20230906_09-42-2 - PA.TSA.AGENT: List all active (running) job lists with their port / pid number:
- to check the last lines of a COMO:
TAIL.COMO tSA_10_20220906_09-42-23 -
Check which jobs lists are currently processed:
LIST F.LOCKING LIKE F.JOB.LIST…
-
Show which jobs are using a specific job list:
CT F.LOCKING F.JOB.LIST.1
F.JOB.LIST.1
001 SGP/SC.BATCH.APP-SC.CONSOL.SWITCH-21 -
Show active processes: SHOW-ITEM-LOCKS (similar to “old” GLOBUS LIST.READU)
-
If need to kill a process or job stuck:
LOGOFF <port/pid>
- nmon -s1 command given via jshell provides also an overview of the CPU and common bin usage.
More docs on this command:
https://www.ibm.com/docs/en/aix/7.2?topic=n-nmon-command - TOPAS again via jshell provides the most active Process ID and system I/O CPU usage and some more useful info about the current status of the system.
Re-launching of a BATCH (COB) process after a crash
After a job crash, selection lists may be still filled with keys. You have then to clear theses files, for EACH COMPANY:
SELECT FBNK.JOB.LIST.1 to 4 then DELETE FBNK.JOB.LIST.1 to 4
Note that 4 is just an example here, it depends on the number of telnet sessions opened during the COB. (Cf field #32 BATCH.SESSIONS from SPF). If you get the following message: “BATCH.JOB.CONTROL -> Aborting” There’s a record ERROR in F.BATCH.STATUS to delete. When it is possible, you can re-run a job which has crashed. If the job is not re-runnable (as T24 tells it you) but you think it can be re-run anyway: Change the property of the job in PGM.FILE in the field 3 ADDITIONAL.INFO to set R (stands for re-runable batch job). Note that you must be 100% sure that this won’t corrupt your COB, for instance there’s not any dependencies with other jobs.
Check a user activity on T24 under Unix
-
- Check the Unix process ID (pid) in the UV column for the enquired user/activity.
Under Jbase: BKS (shell T24):
WHERE (V <pid>
The Command column contains full details.
-
- Check details about a phantom process
You can create a paragraph in your voc with:
ED VOC CHECK.PHANTOM
0001 PQN 0002 IF # %2 MV #1 "ps -eaf|grep -E -i 'PHANTOM|catalog|xrlistener|jbase_agent'" 0003 IF %2 MV #1 "ps -eaf|grep -E -i 'PHANTOM|catalog|xrlistener|jbase_agent'|grep -i '"%2"'" 0004 PU
Example: CHECK.PHANTOM SWIFT will show you all instances of phantom’s names containing SWIFT under your server.
In which library has been compiled a program
ED VOC <prog_name> or jshow -c <prog_name>
Show current locks
show-item-locks or jrla -d V or LIST.READU
Create a standard file under Jbase
create.file FILENAME TYPE=UD
Search in an archived F.JOURNAL under Jbase
LIST.JS 20120201 WITH TXN.REF LIKE DXORD12025…
with LIST.JS is an entry in the VOC declared like this (depending on your history files directory)
0001 PQN 0002 IF # %2 XArgument missing. Usage: LIST.JS YYYYMMDD or list.js YYYYMMDD 0003 MV #1 "LIST ../bnk.arc/F.JOURNAL.SAVE/"*%2*" USING DICT F.JOURNAL " %3,* 0004 P
Resize a table
-
- Let Jbase analysing a table automatically:
jrf -R FBNK.SEC.TRADE
-
- Let Jbase applying its analyse:
jrf -V FBNK.SEC.TRADE
-
- Force a resize with a different modulo/separator (m) (2 notions were distinct under Universe, now same under Jbase):
jrf -V Sm -D FBNK.SEC.TRADE