Cache
By default, batches requests are cached locally.
This is very useful for cases where the long-running process is interrupted:
- user locking computer if running locally
- network error on polling
- Server-side API error during polling
When you re-run a script using batchling, the batch requests belonging to batches that were already sent to remote servers are cached locally.
This ensures you are never invoiced twice or more for a given request.
Whenever a cache hit is made, the corresponding request skips the batch submission part and is directly sent to polling the batch it belonged to in the past, usually giving instant results too.
Cache retention
By default, the cache is located in your home folder under: $HOME/.cache/batchling
Cache is kept locally and automatically cleaned up once it's 30 days old.
This duration aligns with what most Batch APIs share on the duration for which they keep exposing batch results.
Cross-frameworks cache
The cache system is based on request hashing, which means that two requests that are strictly equal will create a cache hit. While possible, it is not guaranteed that all frameworks build requests exactly the same way and thus we cannot guarantee cross-frameworks cache hits.
Deactivating cache
If you ever need to deactivate cache, you can do this through the parameter:
-
cache=Falseif you use the python SDK -
--no-cacheif you use the CLI
Next Steps
- See how dry run can help you plan that everything is ok before sending batches