You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Common theme or question recently is how to get a driver setup that does not require a fully functional ElasticSearch based job registry (EJR), which is not trivial.
The only simpler alternative we have is currently an "in memory" registry, based on an (in process) dictionary. This is fine for unit testing and simple one-off cases but might be too simple and limited for slightly more advanced cases, because there is no persistence.
I was thinking if we couldn't make a SQLite based job registry, which would offer a couple of interesting features:
there sqlite support in the python stdlib, so this would not drag in extra dependencies
standard SQL interface, which hopefully makes the implementation easy and low on boilerplate
persistence without dependency on external service
it both supports purely in-memory databases (e.g. interesting for unit testing) and persistence to disk as a single file (which allows various deployment options: work with local file, work on a docker mount to persist the data outside a container, ...)
The text was updated successfully, but these errors were encountered:
Common theme or question recently is how to get a driver setup that does not require a fully functional ElasticSearch based job registry (EJR), which is not trivial.
The only simpler alternative we have is currently an "in memory" registry, based on an (in process) dictionary. This is fine for unit testing and simple one-off cases but might be too simple and limited for slightly more advanced cases, because there is no persistence.
I was thinking if we couldn't make a SQLite based job registry, which would offer a couple of interesting features:
The text was updated successfully, but these errors were encountered: