MPCO recorder error

[kolozvari]

Hello, I am receiving the following message with my model. I am running model with python API via Powershell since I am running some parametric analysis

I am using the following parameric recorder option from one of the other threads on the forum.

# PARAMETRIC RECORDER:
# this is a trick... when STKO writes the mpco recorder command, it also writes
# a support *.mpco.cdata file with the same name of the recorder. However in STKO the
# recorder name has a suffix = $param_id, that will be evaluated by TCL when creating the
# *.mpco file, but not at the time STKO writes the *.mpco.cdata file. So we can
# simply copy the one created by STKO and rename it accordingly.

# this is the name STKO gave it... make sure to use the escape character before the $ char!!
set STKO_file_name "SSI_\$param_id.part-$PID.mpco.cdata"; # OLD: with OpenSeesMP?
#set STKO_file_name "\$mpco_fname.mpco.cdata"

# this is the new name, DON'T use the escape character, to the param_id will be eval by TCL!
set current_file_name "SSI_$param_id.part-$PID.mpco.cdata"; # OLD: with OpenSeesMP?
#set current_file_name "$mpco_fname.mpco.cdata"

file copy -force $STKO_file_name $current_file_name

this is the error I am getting. Please advise - thanks!

src\H5Fint.c line 2179 in H5F__flush_phase2(): prep for MDC flush failed
major: Object cache
minor: Unable to flush data from cache
#007: C:\J\w\prod\BuildSingleReference\.conan\data\hdf5\1.14.0\_\_\build\f491c24a6c1b13acb3cc68e656f08b3ca13fc643\src\srta\hdf5\1.14.0\_\_\build\f491c24a6c1b13acb3cc68e656f08b3ca13fc643\src\src\H5Fint.c line 2179 in H5F__flush_phase2(): prep for MDC flush failed
major: Object cache
minor: Unable to flush data from cache
#007: C:\J\w\prod\BuildSingleReference\.conan\data\hdf5\1.14.0\_\_\build\f491c24a6c1b13acb3cc68e656f08b3ca13fc643\src\src\H5AC.c line 1230 in H5AC_prep_for_file_flush(): slist enabled failed
major: Object cache
minor: Internal error detected
#008: C:\J\w\prod\BuildSingleReference\.conan\data\hdf5\1.14.0\_\_\build\f491c24a6c1b13acb3cc68e656f08b3ca13fc643\src\src\H5C.c line 2940 in H5C_set_slist_enabled(): slist already enabled?
major: Object cache
minor: Internal error detected
MPCORecorder Error: cannot flush file on record()
b3ca13fc643\src\src\H5C.c line 2940 in H5C_set_slist_enabled(): slist already enabled?
major: Object cache
minor: Internal error detected
MPCORecorder Error: cannot flush file on record()
\w\prod\BuildSingleReference\.conan\data\hdf5\1.14.0\_\_\build\f491c24a6c1b13acb3cc68e656f08b3ca13fc643\src\src\H5AC.c line 1230 in H5AC_prep_for_file_flush(): slist enabled failed
major: Object cache
minor: Internal error detected
#008: C:\J\w\prod\BuildSingleReference\.conan\data\hdf5\1.14.0\_\_\build\f491c24a6c1b13acb3cc68e656f08b3ca13fc643\src\src\H5C.c line 2940 in H5C_set_slist_enabled(): slist already enabled?
major: Object cache
minor: Internal error detected
MPCORecorder Error: cannot flush file on record()
Increment: 9 | Iterations: 6 | Norm: 8.570e-06 | Progress: 90.000 %
\f491c24a6c1b13acb3cc68e656f08b3ca13fc643\src\src\H5AC.c line 1230 in H5AC_prep_for_file_flush(): slist enabled failed
major: Object cache
minor: Internal error detected
#008: C:\J\w\prod\BuildSingleReference\.conan\data\hdf5\1.14.0\_\_\build\f491c24a6c1b13acb3cc68e656f08b3ca13fc643\src\src\H5C.c line 2940 in H5C_set_slist_enabled(): slist already enabled?
major: Object cache
minor: Internal error detec\H5AC.c line 1230 in H5AC_prep_for_file_flush(): slist enabled failed
major: Object cache
minor: Internal error detected
#008: C:\J\w\prod\BuildSingleReference\.conan\data\hdf5\1.14.0\_\_\build\f491c24a6c1b13acb3cc68e656f08b3ca13fc643\src\src\H5C.c line 2940 in H5C_set_slist_enabled(): slist already encted
MPCORecorder Error: cannot flush file on record()
abled?
major: Object cache
minor: Internal error detected
MPCORecorder Error: cannot flush file on record()

[STKO Team]

It should not be a problem.
The new OpenSees (3.7.0) is using an updated version of HDF5 that prints that warning when the system cannot flush the data from memory to the disk.
It's annoying and we will try to make it silent in the future. Since it's in the HDF5 library it may take a while

[kolozvari]

Thanks. Good to know if is not an issue.

[STKO Team]

You're welcome.

Just to explain what it does. Since most recorders in OpenSees do not automatically flush contents to disk on each analyze call, we internally call the HDF5 function to "Try" to force the flush to disk, so that the user is able to visualize results while the analysis is running.

However, flushing data from memory to disk in I/O operation is ultimately decided by the operating system. That's why HDF5 has a function that "tries" to flush, but it's not guaranteed.
The only annoying thing is that in this new version, when it cannot flush, it writes that long warning message...
That's it.

But the behavior is the same as before