Autogen not getting correct number of processes

[Clark Powell]

Actual process count is almost 1500 but autogen says the highest is 325.

MAXPROCESSCNT parameter information:
Feedback information.
Old value was 3000, New value is 3000
Maximum Observed Processes: 325
- AUTOGEN parameter calculation has been overridden.
The calculated value was 2400. The value 3000
will be used in accordance with the following requirements:
MAXPROCESSCNT has been specified by a hard-coded value of 3000.

[Clark Powell]

Sorry, I'm tired today. We're using OpenVMS 8.3 with ECO Update 12.

[Ian Miller. · Accepted solution]

Have you collected feedback information during a peak time when there are 1500 processes?

SYS$SYSTEM:AGEN$FEEEDBACK.EXE writes a text file SYS$SYSTEM:AGEN$FEEDBACK.DAT

which you can search

$ sear sys$system:agen$feedback.dat proc
PROCESSES_PEAK = 78
MAXPROCESSCNT_CUR = 227

[labadie_1]

If you do

$ mc agen$feedback

$ sea sys$system:agen$feedback.dat processcnt_cur

do you see the correct value or not ?

[Clark Powell]

That is exactly the problem but somehow I thought that autogen feedback data was collected automatically. What's the actual method and time?

[Volker Halle]

Clark,

AUTOGEN collects and uses feedback data, if you tell it to.

Autogen saves the feedback data in the SAVPARAMS phase or during SHUTDOWN with SAVE_FEEDBACK.

Autogen uses the feedback data in the GENPARAMS phase unless you specify P3 as NOFEEDBACK or there is some problem with the feedback data.

Check for messages in SYS$SYSTEM:AGEN$PARAMS.REPORT.

If you want AUTOGEN to save the feedback data and use it, specify:

$ @SYS$UPDATE:AUTOGEN SAVPARAMS GENPARAMS

Then check SYS$SYSTEM:AGEN$PARAMS.REPORT. If everything looks o.k., invoke AUTOGEN again with:

$ @SYS$UPDATE:AUTOGEN GETDATA SETPARAMS

and then reboot.

Volker.

[Ian Miller.]

Note doing
@SYS$UPDATE:AUTOGEN SAVPARAMS SAVPARAMS
at a time of peak load is a good idea. Some of the collected data is a snapshot of current data.

You can do

@SYS$UPDATE:AUTOGEN GETDATA TESTFILES

to use the collected feedback to generate a report at another time.

[Clark Powell]

Speaking for the few people with whom I've discussed AUTOGEN, this might be one of the big misconceptions in OpenVMS. I think it's possible that many people assume that the feed back data is being collected automatically, all the time. I like the idea of running at a peak time,
@SYS$UPDATE:AUTOGEN SAVPARAMS SAVPARAMS
and then running AUTOGEN to use the data at a later date.

But even if you collect feedback at a high period, what if you run AUTOGEN at a low useage time and it collects the low useage feedback? Won't that distort the result as well?

[Volker Halle]

Clark,

the PEAK values will still be preserved as long as you don't reboot the system. If you have to reboot after a phase of high usage and before willing to run AUTOGEN, use the SAVE_FEEDBACK option during shutdown.

Volker.

[RBrown_1]

If you do @AUTOGEN SAVPARAMS SAVPARAMS at a busy time and then do the rest of the autogen later then you have to start the second autogen AFTER SAVPARAMS so that you don't collect new feedback data. So

$ @AUTOGEN GETDATA TESTFILES

or

$ @AUTOGEN GETDATA SETPARAMS

etc.