Different behaviour between fds and fds_openmp

[rupole1185]

Describe the bug
SEGMENTATION FAULT

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
libpthread-2.31.s 000014909224D420 Unknown Unknown Unknown
fds 000000000738DED9 Unknown Unknown Unknown
fds 000000000738D170 Unknown Unknown Unknown
fds 0000000006AF79E7 Unknown Unknown Unknown
fds 0000000006AB6BCE Unknown Unknown Unknown
fds 000000000040A71D Unknown Unknown Unknown
libc-2.31.so 000014909030E083 __libc_start_main Unknown Unknown
fds 000000000040A636 Unknown Unknown Unknown

= BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
= RANK 0 PID 11757 RUNNING AT hpc-server49403
= KILLED BY SIGNAL: 9 (Killed)

To Reproduce
Use the attached file [42122-sc6-A-02.fds]
Run FDS6.9.1 on 60 cores single thread ( mpiexec -np 60 fds file_name.fds )
The issue seems to be related to the high number of files opened by FDS. I have also tried to increase the limit of files using the following: ulimit -n 99999

But I still got the error.

If instead you execute the very same case using 2 threads ( mpiexec -np 60 fds_openmp file_name.fds ) the simulation works ok.

Expected behavior
Simulation should work both when single and when multi-threads.

Desktop (please complete the following information):

• OS: Ubuntu
• Version: 2004

42122-sc6-A-02.fds.txt

[drjfloyd]

Fortran limits the number of open files per process but I don't think your case is coming close to that limit as the open files would be the error, out, and csv files which are just a handful of files. SLCF is opened and closed per write. When does your case fail? I am running now with 6.9.1 and it is out to over 2 s.

[rupole1185]

Hi, the failure is right at the beginning. And I only got it when running with the configurations I said, so when running without hyperthread Ruggero Poletto Managing director VideoCall: calendly.com/ruggero-poletto | Mobile: +39 378 30 33 133 Il mar 23 lug 2024, 13:07 Jason Floyd ***@***.***> ha scritto:

…

Fortran limits the number of open files per process but I don't think your case is coming close to that limit as the open files would be the error, out, and csv files which are just a handful of files. SLCF is opened and closed per write. When does your case fail? I am running now with 6.9.1 and it is out to over 2 s. — Reply to this email directly, view it on GitHub <#13207 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABXMUCLJGHVUL3VSIPCVINLZNY2QHAVCNFSM6AAAAABLKEOUMSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENBUHEZTONBWGE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[drjfloyd]

Are you running the precompiled binaries or have you compiled on your own?

[rupole1185]

The precompiled binaries Ruggero Poletto Managing director VideoCall: calendly.com/ruggero-poletto | Mobile: +39 378 30 33 133 Il mar 23 lug 2024, 13:23 Jason Floyd ***@***.***> ha scritto:

…

Are you running the precompiled binaries or have you compiled on your own? — Reply to this email directly, view it on GitHub <#13207 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABXMUCI5FHBGSJSTFNMBQO3ZNY4KBAVCNFSM6AAAAABLKEOUMSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENBUHE3TCOBQGE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[drjfloyd]

This is running without issues for me. What does the start of the .out file say for the case that fails?

Fire Dynamics Simulator

Current Date : July 23, 2024 07:04:18
Revision : FDS-6.9.1-0-g9743202-release
Revision Date : Fri Apr 5 16:53:17 2024 -0400
Compiler : Intel(R) Fortran Intel(R) 64 Compiler Classic for applications running on Intel(R) 64, Version 2021.7.1 Build 20221019_000000
Compilation Date : Apr 06, 2024 04:35:02

Number of MPI Processes: 60

[rupole1185]

Fire Dynamics Simulator Current Date : July 23, 2024 08:01:45 Revision : FDS-6.9.1-0-g889da6a-release Revision Date : Sun Apr 7 17:05:06 2024 -0400 Compiler : Intel(R) Fortran Intel(R) 64 Compiler Classic for applications running on Intel(R) 64, Version 2021.7.1 Build 20221019_000000 Compilation Date : Apr 09, 2024 13:24:51 Number of MPI Processes: 60 *Ruggero Poletto - **Managing director* *VideoCall*: calendly.com/ruggero-poletto | *Mobile*: +39 378 30 33 133 *NEXT EVENTS we are attending:* FEMTC <https://www.femtc.com/events/2024> - 18-20 Sept. 2024 | OPENFOAM ITA - 7-8 Sept. 2024

…

________________________________________ CFD FEA SERVICE SRL P.IVA 04545570238 - via Borgo Grande 19, 37044 Cologna Veneta VR *ATTENTION:* All information in this e-mail message, as well as any files included, are intended for the exclusive use of the recipient (s) indicated above / indicated, and therefore subject to protection pursuant to law no. 675 and of the Legislative Decree of 30 June 2003 n. 196. Anyone who receives this message by mistake, is requested to notify the writer of the incorrect receipt, and proceed with the immediate cancellation of the same from the system. Your cooperation is appreciated. *PRIVACY:* All personal data, including e-mail addresses, are collected and stored with the sole purpose of carrying out our professional activity towards you. The privacy law D.Lgs. N. 196 of 30/06/2003 provides in article 7 the right of the interested party to access their personal data, to request their correction, updating and deletion, therefore we invite you to report us in the way you deem most appropriate as provided by Article 7.

On Tue, 23 Jul 2024 at 13:25, Jason Floyd ***@***.***> wrote: This is running without issues for me. What does the start of the .out file say for the case that fails? Fire Dynamics Simulator Current Date : July 23, 2024 07:04:18 Revision : FDS-6.9.1-0-g9743202-release Revision Date : Fri Apr 5 16:53:17 2024 -0400 Compiler : Intel(R) Fortran Intel(R) 64 Compiler Classic for applications running on Intel(R) 64, Version 2021.7.1 Build 20221019_000000 Compilation Date : Apr 06, 2024 04:35:02 Number of MPI Processes: 60 — Reply to this email directly, view it on GitHub <#13207 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABXMUCORCNGTPMIO4OWZYVDZNY4Q5AVCNFSM6AAAAABLKEOUMSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENBUHE3TIOJZHE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[drjfloyd]

@mcgratta @rmcdermo @gforney Does this case run for you? I can't reproduce the error on our system. It is running for mew with release and the current source.

[rmcdermo]

The latest compiled source ran fine for me. But the 6.9.1 release failed, see test.err.txt below.
test.err.txt
test.fds.txt

[drjfloyd]

@rupole1185 can you try with the latest test release and see if that works for you?

https://github.com/firemodels/test_bundles

[rupole1185]

Yeah. I'll try it tomorrow if it is not an issue. Thanks Ruggero Poletto Managing director VideoCall: calendly.com/ruggero-poletto | Mobile: +39 378 30 33 133 Il mar 23 lug 2024, 15:15 Jason Floyd ***@***.***> ha scritto:

…

@rupole1185 <https://github.com/rupole1185> can you try with the latest test release and see if that works for you? https://github.com/firemodels/test_bundles — Reply to this email directly, view it on GitHub <#13207 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABXMUCJGUAX3C2UFM22R7PTZNZJQPAVCNFSM6AAAAABLKEOUMSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENBVGIZDIOBYGE> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[rupole1185]

Hi there, I confirm that when using the most recent version of FDS (FDS-6.9.1-594-gdec1605 2024-Jul-26 15:35) the problem has been solved. Thanks, *Ruggero Poletto - **Managing director* *VideoCall*: calendly.com/ruggero-poletto | *Mobile*: +39 378 30 33 133 *NEXT EVENTS we are attending:* FEMTC <https://www.femtc.com/events/2024> - 18-20 Sept. 2024 | OPENFOAM ITA - 7-8 Sept. 2024

…

________________________________________ CFD FEA SERVICE SRL P.IVA 04545570238 - via Borgo Grande 19, 37044 Cologna Veneta VR *ATTENTION:* All information in this e-mail message, as well as any files included, are intended for the exclusive use of the recipient (s) indicated above / indicated, and therefore subject to protection pursuant to law no. 675 and of the Legislative Decree of 30 June 2003 n. 196. Anyone who receives this message by mistake, is requested to notify the writer of the incorrect receipt, and proceed with the immediate cancellation of the same from the system. Your cooperation is appreciated. *PRIVACY:* All personal data, including e-mail addresses, are collected and stored with the sole purpose of carrying out our professional activity towards you. The privacy law D.Lgs. N. 196 of 30/06/2003 provides in article 7 the right of the interested party to access their personal data, to request their correction, updating and deletion, therefore we invite you to report us in the way you deem most appropriate as provided by Article 7.

On Tue, 23 Jul 2024 at 15:16, Ruggero Poletto @ CFD FEA SERVICE < ***@***.***> wrote: Yeah. I'll try it tomorrow if it is not an issue. Thanks Ruggero Poletto Managing director VideoCall: calendly.com/ruggero-poletto | Mobile: +39 378 30 33 133 Il mar 23 lug 2024, 15:15 Jason Floyd ***@***.***> ha scritto: > @rupole1185 <https://github.com/rupole1185> can you try with the latest > test release and see if that works for you? > > https://github.com/firemodels/test_bundles > > — > Reply to this email directly, view it on GitHub > <#13207 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/ABXMUCJGUAX3C2UFM22R7PTZNZJQPAVCNFSM6AAAAABLKEOUMSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDENBVGIZDIOBYGE> > . > You are receiving this because you were mentioned.Message ID: > ***@***.***> >

[rmcdermo]

Thanks for letting us know.