In some use cases, avoiding parsing perf data for multiple times contributes the flamescope server performance & its users' experience. flamescope already has an in-memory cache for the analyzed perf data, and this PR adds file-backed one in case of in-memory cache miss.
In our case, we render some graphs in a page by querying to flamescope which we run with three or a few more instances and the most simplest way to share processed data between them is file-backed cache.
To enable this feature, change USE_HEATMAP_FILECACHE in app/config.py to True.
In some use cases, avoiding parsing perf data for multiple times contributes the flamescope server performance & its users' experience. flamescope already has an in-memory cache for the analyzed perf data, and this PR adds file-backed one in case of in-memory cache miss. In our case, we render some graphs in a page by querying to flamescope which we run with three or a few more instances and the most simplest way to share processed data between them is file-backed cache.
To enable this feature, change
USE_HEATMAP_FILECACHE
in app/config.py toTrue
.Related issue: https://github.com/Netflix/flamescope/issues/58