I am trying to access my simulation data in the cluster using the pvserver and the connection becomes unstable after I select the nek5000 field file to open and displays the message below.
Can you explain what is happening and the rectification?
I can open this file using the normal local paraview, just that the pvserver version has this issue. I will open an issue.
PS: the paraview version I am using is 5.12.0
Hi! when I opened the nek5000 file in a notepad, it looked like a metadata file that points to other files. I don’t know if I am wrong but without the actual linked data, I was wondering how ParaView/pvserver can open it. When I opened the file in local ParaView, obviously I get the same error you are getting. Since you have not shared the nek5000 linked files with Mathieu, I wish to ask… did you miss to copy the nek5000 linked files to the cluster as well?
This thread also reminds me of the [*.mhd and .raw file] pair, where .mhd is a metadata pointing to a .raw data. Failing to provide the raw data in the same folder while loading the .mhd file is a problem, but there I usually get an error instead of a crash.
Yes, as I mentioned before this is a binder file that connects other field files which has the timestamp data. I just posted the binder file and my cluster had the linked data. Each file has around 2.3 GB of data and hence I couldn’t post the data here.
Just to add to that, I think there is a problem in the cluster version of paraview as when I try to open it directly by loading paraview in the cluster I get the following error:
Thanks for showing the logs. Your cluster version of ParaView is not too old.
Without having the linked files, I get the following error messages (local ParaView 5.12.0) before the crash:
Error parsing file. Unknown tag type:
( 6.956s) [paraview ] loguru.cpp:479 1| . atexit
[sherin-pc:10063] *** Process received signal ***
[sherin-pc:10063] Signal: Segmentation fault (11)
[sherin-pc:10063] Signal code: Address not mapped (1)
[sherin-pc:10063] Failing at address: 0x98c
I will open an issue.
Can I know why the paraview installed on cluster is failing whereas if I try to open the file with local, though it keeps on displaying the message of not responding… eventually I get the simulation displayed albeit being very slow. Is it because of mpi distribution in the cluster?
Can I avert this issue with the new version of 5.13?
As I said the complete data is large, so I have few timestamps shared with you. This problem is in the cluster version(mpi) and not in the local (serial - RC1) version.
I am using the VisIt Nek5000 reader. As I said I could open this using a local paraview via remote folders to the cluster whereas if I open it directly using the paraview in the cluster I get the message as discussed above. Nek5000 reader never worked and resulted in a crash always
Can I know the reason why you cannot open the file locally? I have managed to open it and then @sherin.sugathan managed to open it as well. Can you be more specific?
ParaView local (5.12.0 prebuilt) opens the file successfully.
ParaView local (5.12.0 custom built from source) fails to read the file.
ParaView remote (5.12.0 custom built from source) when accessed directly, fails to read the file.
I may be wrong, but I am just making inferences based on my above observations. I suspect that the cluster version of your ParaView was built from source and without the necessary CMAKE option needed for enabling the VisIt nek reader.
Earlier, you said that you can run ParaView directly in your cluster. Could you please run ParaView directly in your cluster, then goto Edit → Settings and then under the IO tab you will find a list named Reader Selection. Scroll down and tell how many entries of “Nek5000 Files” you see there. 1 or 2?