scr
scr copied to clipboard
LLNL
SCR caches checkpoint data in storage on the compute nodes of a Linux cluster to provide a fast, scalable checkpoint / restart capability for MPI codes.
SCR can direct the application to write dataset files to subdirectories within a cache directory. SCR also stores its redundancy data in these subdirectories. **Question**: Should it be considered an...
The SCR library communicates tiwh the nnf-dm server via a socket file name. We need do create a settable `SCR_CONFIG_AXL_NNFDM`/`AXL_KEY_CONFIG_NNFDM` that may be provided via environment and/or SCR_Config call.
We should consider: https://exaworks.org/psi-j-python/ It’s a portable submission interface for job schedulers in python. Maybe this will be helpful in SCR.
SCR is getting large enough that importing is getting cumbersome. For example, updating to v0.3.0 just required this entire file: https://github.com/LLNL/axom/blob/develop/src%2Fcmake%2Fthirdparty%2FFindSCR.cmake As well as a lot of extra directories in...
The scavenge operation assumes files are cached in node local storage. After the run stops, the scavenge script launches the scr_copy executable on each compute node via pdsh. On each...
From Adam: With the current version of AXL, we'd have a single thread on a single compute node copy the entire file from cache to the parallel file system. However,...
When using a global file system as cache, we need to examine the number of metadata files that SCR creates for each dataset. SCR stores a filemap for each process...