Hello,
I had a few questions about using recorders with OpenSeesMP. I heard somewhere that -file recorders do not work well in OpenSeesMP, and that -xml recorders should be used instead. Most of my post-processing codes were designed for -file recorders, so it is a bit tough to look at the data.
However, the bigger issue right now is that the -xml recorders do not appear to start at the beginning of the analysis. For instance, I am running analyses of a building under GM. I start with setTime 4.0 because I want to start 4.0 seconds into the GM record. But the data in my -xml files start at weird times, anywhere from 4.0 to 4.88. Additionally, there are steps of gravity load before the GM that the recorders are also missing.
Do -xml files take some time to "start up", or what could be causing them to miss data at the beginning? I'm not sure if this is specifically an OpenSeesMP issue or a -xml recorder issue, as this is my first time using both.
xml recorder using OpenSeesMP
Moderator: selimgunay
Re: xml recorder using OpenSeesMP
As an update, this issue is still unresolved, and DOES appear to be an issue with hybrid analysis. I attempted both -file recorders and -xml recorders, and both have a similar issue. Several recorders have the full data set, but many of them have an incomplete first line of data, and that data is not necessarily associated with the 1st analysis step. Some recorder files are starting dozens of steps into the analysis before printing any data into the file. I tried inserting many 'barrier' commands (even including one after every recorder definition) to make sure parts of the program were not proceeding before other processes were ready. I also added many 'after' commands to force the analysis to pause, but neither of these changes resolved the issue. As I said, this is occuring both with -file recorders and with -xml recorders, but only when using OpenSeesMP.
Re: xml recorder using OpenSeesMP
the recorders in MP work fine for all the scripts i have tested. one possible problem source is that the recorders are writing to the same file. that doesn't work and it is not supposed too.