version 1.9.1 vs version 2.0

[epeyg...@gmail.com]

Dear OpenQuake support team ,

I am running a scenario risk analysis regarding Hayward Rodgers fault with Abrahamson silva 2008 as GMPE.

I have been able to run it before with version 1.9.1; however, with new updates to version 2, I am not able to run it any more.

I put the log files and job file for your information. Is this a change in model capacity or I have to make the job file in different way?

Thanks,

Elnaz

[epeyg...@gmail.com]

I would like to add if I select smaller number of ruptures for example 2000 instead of 10000 if will not face with such error.

Thanks,
Elnaz

[Daniele Viganò]

Dear Elnaz,

to better understand your issue we need some more information:

1. It seems to me that you are running the oq-engine inside one of our VBox VMs, right?
2. Please provide the output of the following command
   dpkg -l | grep python-oq
3. Just after the segfault happens please provide the output from the following command:
   dmesg | grep seg
4. Have you tried to upgrade OQ to v2.1 and running the job again?
   You can upgrade running apt update && apt install python-oq-engine
   If you are running a VM make a snapshot first, so you can easily revert to the previous state.

Cheers,
Daniele

[epeyg...@gmail.com]

Dear Daniele,

1.yes
2.
openquake@oq-1404:~$ dpkg -l | grep python-oq
ii  python-oq-engine                           2.0.0-0~trusty01                        all          computes seismic hazard and physical risk
ii  python-oq-hazardlib                        0.20.0-0~trusty01                       amd64        hazardlib is a library for performing seismic hazard analysis

3.
openquake@oq-1404:~$ dmesg | grep seg
[ 1456.441536] oq[2235]: segfault at 24 ip 00000000005377c7 sp 00007fff9d8072d0 error 6 in python2.7[400000+2bc000]
[ 6240.369828] oq[3827]: segfault at 24 ip 00000000005377c7 sp 00007ffea1073b40 error 6 in python2.7[400000+2bc000]
 4.
I have done the update righ now and getting same error.
here is the result :
openquake@oq-1404:~$ oq-engine --run /media/sf_Share_folder_with_Linux/test_for_vahid/job_hazard_risk.ini
The command 'oq-engine' is deprecated and will be removed. Please use 'oq engine' instead
[2016-10-20 21:13:19,480 #6 INFO] Using engine version 2.1.0
[2016-10-20 21:13:19,531 #6 INFO] Using hazardlib version 0.21.0
[2016-10-20 21:13:19,585 #6 INFO] Reading the exposure
[2016-10-20 21:13:44,428 #6 INFO] Read 43750 assets
[2016-10-20 21:13:44,750 #6 WARNING] Please upgrade /media/sf_Share_folder_with_Linux/test_for_vahid/vuln_model.xml to NRML 0.5
[2016-10-20 21:14:16,118 #6 INFO] Building the epsilons
[2016-10-20 21:14:16,260 #6 INFO] Building (2917, 10000) epsilons for taxonomy C1L
[2016-10-20 21:14:18,036 #6 INFO] Building (2917, 10000) epsilons for taxonomy C2L
[2016-10-20 21:14:20,031 #6 INFO] Building (2917, 10000) epsilons for taxonomy C3L
[2016-10-20 21:14:21,626 #6 INFO] Building (2917, 10000) epsilons for taxonomy MH
[2016-10-20 21:14:23,203 #6 INFO] Building (2917, 10000) epsilons for taxonomy PC1
[2016-10-20 21:14:24,868 #6 INFO] Building (2917, 10000) epsilons for taxonomy PC2L
[2016-10-20 21:14:26,414 #6 INFO] Building (2917, 10000) epsilons for taxonomy RM1L
[2016-10-20 21:14:28,075 #6 INFO] Building (2917, 10000) epsilons for taxonomy RM2L
[2016-10-20 21:14:29,735 #6 INFO] Building (2917, 10000) epsilons for taxonomy S1L
[2016-10-20 21:14:31,296 #6 INFO] Building (2917, 10000) epsilons for taxonomy S2L
[2016-10-20 21:14:32,862 #6 INFO] Building (2917, 10000) epsilons for taxonomy S3
[2016-10-20 21:14:34,462 #6 INFO] Building (2917, 10000) epsilons for taxonomy S4L
[2016-10-20 21:14:36,039 #6 INFO] Building (2917, 10000) epsilons for taxonomy URML
[2016-10-20 21:14:37,599 #6 INFO] Building (2913, 10000) epsilons for taxonomy W1
[2016-10-20 21:14:39,149 #6 INFO] Building (2916, 10000) epsilons for taxonomy W2
/usr/bin/oq-engine: line 39:  5055 Killed                  $OQ_ENTRY engine $@

I again provide the out put of the asked commands :

openquake@oq-1404:~$ dpkg -l | grep python-oq
ii  python-oq-engine                           2.1.0-0~trusty01                        all          computes seismic hazard and physical risk
ii  python-oq-hazardlib                        0.21.0-0~trusty01                       amd64        hazardlib is a library for performing seismic hazard analysis

openquake@oq-1404:~$ dmesg | grep seg
[ 1456.441536] oq[2235]: segfault at 24 ip 00000000005377c7 sp 00007fff9d8072d0 error 6 in python2.7[400000+2bc000]
[ 6240.369828] oq[3827]: segfault at 24 ip 00000000005377c7 sp 00007ffea1073b40 error 6 in python2.7[400000+2bc000]

Thanks
Elnaz

[Daniele Viganò]

Dear Elnaz,

it would be possible for you to send us (at engine....@openquake.org) a zip file containing the full set of input files? The sole job.ini has not been enough to exactly reproduce your issue, since it seems to be triggered only in a very specific occasion.

Thank you

Daniele

--
You received this message because you are subscribed to the Google Groups "OpenQuake Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openquake-use...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

DANIELE VIGANÒ | System Administrator | Skype dennyv85 | +39 0382 5169882

GLOBAL EARTHQUAKE MODEL | working together to assess risk

GEM - globalquakemodel.org | T - @GEMwrld | F - GEMwrld

[epeyg...@gmail.com]

 Dear Daniele,

I have shared the input data regarding the analysis with you in the following email.

Thanks,
Elnaz

To unsubscribe from this group and stop receiving emails from it, send an email to openquake-users+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Daniele Viganò]

Dear Elnaz,

thanks, we have received them.

One last question: how much memory and how many cores have you allocated to the VM?

Thanks,
Daniele

To unsubscribe from this group and stop receiving emails from it, send an email to openquake-use...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Michele Simionato]

On Wednesday, October 26, 2016 at 4:23:38 PM UTC+2, epeyg...@gmail.com wrote:

 Dear Daniele,

I have shared the input data regarding the analysis with you in the following email

Thanks for sharing. I have analyzed the problem and I know how to solve it. You should expect the next version of the engine to be a lot more efficient for this kind of computation.

For the time being, the scientific stuff tells me that number_of_ground_motion_fields=10000 is quite large. You can reduce it to 1000 and have a computation 10 times faster.

HTH,

      Michele 

[epeyg...@gmail.com]

Dear Michele/ Daniele,
Thanks for your response and your perfect support.

Daniele,
I have dedicated 14 Gb ram and I played with 1 to 6 core to this analysis.
 
Michele,

I performed this analysis many times with version 1.9 OpenQuake and it worked more efficient very and fast; however, in there I could not use the Abrahamson 2014 which you provide it in new version 2.1. Now in this version 2.1 I can not run even the model for Abrahamson 2008. Unfortunately, I need to perform same analysis to compare the results because the results with 1000 and 10000 simulation are very different as I have performed sensitivity analysis (According to the accuracy I need). would you please let me know your estimation in time when the new version of OpenQuake solving this issue would be available? Please if it is possible let me know if you have any estimation how much memory and core this analysis need?
I appreciate your help and support.

Thanks,
Elnaz

[Michele Simionato]

Il giorno giovedì 27 ottobre 2016 17:15:36 UTC+2, epeyg...@gmail.com ha scritto:

Dear Michele/ Daniele,
Thanks for your response and your perfect support.

Daniele,
I have dedicated 14 Gb ram and I played with 1 to 6 core to this analysis.
 
Michele,

I performed this analysis many times with version 1.9 OpenQuake and it worked more efficient very and fast; however, in there I could not use the Abrahamson 2014 which you provide it in new version 2.1. Now in this version 2.1 I can not run even the model for Abrahamson 2008. Unfortunately, I need to perform same analysis to compare the results because the results with 1000 and 10000 simulation are very different as I have performed sensitivity analysis (According to the accuracy I need). would you please let me know your estimation in time when the new version of OpenQuake solving this issue would be available? Please if it is possible let me know if you have any estimation how much memory and core this analysis need?
I appreciate your help and support.

I do not know when the engine version 2.2 will be released. The problem you discovered does not warrant a new release because it is just a performance bug: the engine is giving correct results, it is only slow.
The bug is already fixed in master, so you can just use the development version and have the fix now or the nightly builds and have the fix tomorrow. The memory occupation should be reduced to half of what it was before. On my workstation the runtime went down from 55 minutes to 1.5 minutes, this is the kind of improvement you should expect, i.e. a lot better than version 1.9.


       Michele