Deploys take way too long. aws s3 performs a full scan to figure out what has changed. Then the actual transfers must take place, possibly consisting of many files. And this all happens in a blocking fashion, once the changes are ready, as opposed to work being done in the background before it is needed.
This is a risk (can't push e.g. an urgent bugfix as quickly), a waste of time (ever wait for a deploy before getting back to work?), and encourages bad habits (avoiding the slow step whenever possible).
"Sync" sets out to improve the situation in whatever ways are possible.
- More advanced planning. Currently only fan-out is supported.
On primary:
sync-primary inst1 <32 random chars> --replica=123.45.67.89:1234 --exclude=^.git
On client:
sync-replica inst2 <32 random chars> 123.45.67.89 1234
NOT SUPPORTED YET: Then after making file changes:
syncctl flush To ensure that changes have been propagated to replicas.