pengubco / guiid Goto Github PK

There are cases where the application would like to get X, e.g. 100, 200, at a time. The Snowflak-ID supports 4K IDs in one milliseconds, so this is an easy and should not incur noticeable longer blocking compared to request of a single ID.

The readme does not have details of how the ID service work. Include this info in a a separate design.md.

Create a client module for customers that talk with the snowflake-id service.

Right now, when a worker starts, it iterates through all possible worker IDs, starting from 0, all the way up to the max worker.
If the only available worker id is 1023. Iterating from 0 to 1022 takes time and are unnecessary.

Fix: list used worker IDs from etcd. Then only iterate through worker IDs that are not returned from etcd.

Across time and across workers, the timestamp associated with a worker ID should always be monotonic increasing. Resetting it back to 0 is wrong.