aiidalab / aiidalab/aiidalab-qe
Automated code install consumes a lot of RAM (up to 8 GB)
- Lenguaje dominante
- Python
- Estrellas
- 33
- Forks
- 25
- Merge medio
- 6 h 7 min
- PR fusionados (30 d)
- 8
Descripción
**Describe the bug**
When the app is opened for the first time, it installs the necessary QE codes locally. This results in a very high memory consumption.
**To Reproduce**
Steps to reproduce the behavior:
1. Install the latest version of the app on an empty container.
2. Open the app (`New calculation`)
3. Monitor the memory consumption with something like this:
```bash
while true; do echo `cat /sys/fs/cgroup/memory.current` / 1024^3 | bc -l ; sleep 1 ;done
```
4. The memory should go up, from ~1.5 GB to ~8 GB.
**Expected behavior**
Memory consumption should stay below 4 GB; otherwise, many users' accounts will unexpectedly crash due to "unknown" problems.
**Some investigation details**
For the first time, we thought it was due to the pymatgen version, but it was a false positive. Together with @cpignedoli, we are still investigating the origin of the problem.
After we increased the pod memory to 8 GB, the code installation process went through, so it seems the problem is not the memory leak (as we thought before) but just a very high memory consumption. See the comment of @cpignedoli below.
Guía de contribución
No hay ninguna guía de contribución indexada para este repositorio
Evaluación
Este issue todavía no se ha evaluado.