DTIO, a new, distributed, scalable, and adaptive I/O System. DTIO is a task-based I/O System, it is fully decoupled, and is intended to grow in the intersection of HPC and BigData.
You should be able to intercept using LD_PRELOAD, make sure to set the environment variable DTIO_CONF_PATH to the path of your DTIO configuration file (e.g., /home/$USER/DTIO/conf/default.yaml), otherwise we won't be able to find it.
Debugging DTIO is a complicated task, please read DTIO/docs/debugging.md if you're interested in contributing to this project.
- Task dependencies (don't care)
- Metadata persistent store (flush at the end)
- Automated server bootstrapping
- Investigate read simulation
- Handle MDM for outstanding operations (data in transit)
- Discuss how important is this issue of decoupled components
- Possible fixes:
- Invalidation lists + timer expiration to clean up
- Intermediate state of data in MDM
- Timeout task scheduling. line#66
- check the
usleep
in task scheduler → infinite looping - Aggregating logs
cat ts_* >> ts.csv
- Printing correctly
std::stringstream stream; // #include <sstream> for this stream << 1 << 2 << 3; std::cout << stream.str();