Still doing full table scan on startup

[kof]

[kof]

It looks like we have to live with this overhead on startup or stop using tailable cursors at all, but instead do the polling using _id.

Current implementation is a mix of using a specific document id and tailable cursors.

Ideally we need a way to start tailable cursor at the latest document and call .nextObject from there, but I don't see how to solve this.

Anybody?

[zhaoyou]

In my case: doing part table scan on startup(version 1.0.4)

[juuxstar]

If you're still experiencing this issue, do a .findOne({}, { sort : { $natural : -1 }}) to find the last record, get it's _id, and start tailing at or after that _id.