Comment 11 for bug 218388

Jamu, this branch doesn't fix the issue brought up in this bug. In fact, it probably makes it even worse, since it's more expensive to insert new rows if there's an index to maintain. The issue described in this bug is somewhat of an edge case which needs a fix on the client side, since it should never be possible to queue so many individual free space messages in a single message.

Specifically for the branch, it looks good. Only one small detail, and a question related to the above:

[1]

+CREATE INDEX custom_graph_data_point_custom_graph_data_id_idx
+ ON custom_graph_data_point (custom_graph_id)

The naming pattern seems to be off here (notice the second _data_). There are two instances of this.

[2]

I'm hoping it won't be a big hit on the database to insert the several data points with the new indexes, but we should keep our eyes open about the impact of this change on the deployment.