Flame elastic scaling

How does elastic scaling work on flame runners?

I have a pool, and I make a few FLAME.cast calls. If the config is set to scale to zero, I always get one runner spawned, but I would expect a new runner booted up for each call. Did I misunderstand something?

My Pool config:

name: PhoenixAlbums.ImageProcessor,
min: 0,
max: 10,
max_concurrency: 1,
single_use: true,
log: :debug

Also is there some kind of caching going on with the runners? I noticed that initially the startup time was ~6s on now it’s down to ~2s. Not a bad change, but I suspect some kind of reuse and it might be obscuring the debugging I’m currently working on. If this is the case, is there a way to empty this cache or configure it to always build new machines?

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

This was a bug in FLAME that was fixed in 0.1.9.

1 Like