Suppress output by Cross-CorrelationOp when run in GPT

snap
#1

Hi developers,

we are running (many) graph files via the GPT in an automated fashion (orchestrated by an external tool). We currently experience the issue that GRAPH’s, containing a Cross-Correlation operator, generate so much output to the stdout, that our logging framework gets overloaded and chokes on it. Is there a way that we can control/reduce the amount of msg this operator produces? It seems that only this specific operator generates so many msg…

Thanks alot in advance!
Sven.

Logging in ESD Operator
#2

Hi Sven,

This is one operator of the S1TBX.
I’ve looked at the code and there are not so many lines which can cause this.
I guess you get output like:

mGCP = ({}, {})123123
Initial sGCP = ({}, {})456456

and

Final sGCP = ({},{})789890
Final sGCP coherence = {}1.2345

I think it should be no problem to remove them. or set the log level to debug.
But @lveci or @junlu should have a look at it.

cheers
Marco

#3

You can try to increase the log level to a higher level. So that the messages are not logged any more.
Try specifying -dsnap.log.level=WARNING with you gpt command.

#4

Hi Marco,

please see the snippet of msg below. I think I would still want the warning to show but get rid of the “INFO:” lines.

I will try changing the log to a higher level.

Thanks,
Sven.

…
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: Coarse sGCP = ({}, {})-1.00.0
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: Coarse sGCP coherence = {}0.42809484868701936
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: mGCP = ({}, {})13888.032406.0
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: Initial sGCP = ({}, {})13887.032406.0
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Using de-meaned magnitude patches for incoherent cross-correlation
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Pixel level offset: -58.0, -55.0 (corr=0.0754140300444644)
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: Coarse sGCP = ({}, {})-55.0-58.0
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: Coarse sGCP coherence = {}0.0754140300444644
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: mGCP = ({}, {})15740.032406.0
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: Initial sGCP = ({}, {})15685.032348.0
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: crosscorrelate:changing static block to size [64, 64]
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Using de-meaned magnitude patches for incoherent cross-correlation
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Pixel level offset: 1.0, 0.0 (corr=0.45127771056242144)
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Careful, decrease AccL or increase winsizeL
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: crosscorrelate:changing static block to size [64, 64]
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Pixel level offset: -15.0, -8.0 (corr=0.2423844403723709)
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Careful, decrease AccL or increase winsizeL
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Careful, decrease AccP or increase winsizeP
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Pixel level offset: 0.0, -4.0 (corr=0.33251294585869995)
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: Coarse sGCP = ({}, {})-4.00.0
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: Coarse sGCP coherence = {}0.33251294585869995
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: mGCP = ({}, {})10184.032406.0
INFO: org.esa.s1tbx.insar.gpf.coregistration.CrossCorrelationOp: Initial sGCP = ({}, {})10180.032406.0
INFO: org.jlinda.core.coregistration.utils.CoregistrationUtils: Using de-meaned magnitude patches for incoherent cross-correlation
…

#5

Also the logging in the CoregistrationUtils class should be reduced. This is also a lot. This might be interesting when debugging but I think not when simply processing data.
You are right. Warnings should still be printed.

#6

Hi Marco,

Thanks for the suggestion. We tried running the gpt command with that option but we received:

Error: Unknown option ‘-dsnap.log.level=WARNING’

I also see in the option that it refers to snap, and not gpt.
Any other suggestion form you? Thanks.

#7

I have to check why this is not working.
Maybe the generic java property is working:
-Djava.util.logging.ConsoleHandler.level=WARNING

#8

Thanks. That option didn’t result in “unknown option error” in a first attempt.
We will test this further.

#9

Sorry about that Sven. I’ll knock it off.

#10

OK, the -Dsnap.log.level=WARNING option must be set early. So it must be included in the gpt.vmoptions file. And it must be a capital ‘D’

#11

Thanks! Let me know when a fix is available. BTW are you back from your Europe trip yet and will possibly anyone of your team attend the “Bringing space to earth” meeting @ The Hague end of this months?

#12

Not yet, just catching up on emails in Frankfurt airport :airplane:
Sorry, I don’t have any more conferences/workshops planned anytime soon. It would be good to meet if I ever pass by ESTEC.

#13

Be sure to do just that! Have a good trip home!