OpenSees SP problem

[Joesh]

Hi all. When I run my model using np = 1, it works fine. When I run my model using np = 2, I get all the way through my script until the part where I analyze the frame under gravity load (command analyze 20) and then I get the message

job aborted:
rank: node: exit code[: error message]
0: JFH-JSS: 123
1: JFH-JSS: 123
2: JFH-JSS: -1073741819: process 2 exited without calling finalize
3: JFH-JSS: 123

My model is too large to post here likely.

Here is my analysis procedure for gravity load portion

constraints Plain
numberer RCM
system Mumps
test NormDispIncr 1.0e-6 10
algorithm Newton
integrator LoadControl 0.05
analysis Static
analyze 20

The types of materials I am using in the model are:
elastic
hysteretic
minmax
elasticPPgap
Steel02
Parallel

The types of elements I am using in the model are:
forceBeamColumn
elasticBeamColumn
zeroLength

I am wondering if the root of the problem is stemming from one of the material types or something. Otherwise why would the model run with np = 1.0, and also the script portion of the model still run with np =2,3,4 - except stop right at the analyze command.

[fmk]

its failing in send/recvSelf .. these methods are not called when np =1. it looks like all the methods have been implemented for the elements and materials you are showing. can you try creating a real simple small model (portal frame) and get the same error to occur.

[zanakarimi]

Hi there,
I get the same problem with sending the job when I use different processors in OpenSeesSP. I don't have this problem when I use 2 or 4 processors but I cannot use 3 or 5 or 6 processors. Should I follow some method of discretization to let the software decompose?
Thanks in advance,