Piotr Grabowski
Piotr Grabowski
After some additional private discussion, for now skipping UDT columns will be the workaround and we agree that adding this new feature cannot be achieved by a quick fix. Nevertheless,...
This PR is also available at scylladb/gocql branch: heavy-loaded-optim and at scylladb/gocql tag: heavy-loaded-optim-v1.
What would be the test scenario? A single overloaded shard (for example due to compaction)? Or someone else would do the actual benchmarking after we implemented it?
An update on the issue: @Gor027 will start with reproducing the issue (in a clean way!) - a very similar scenario @vladzcloudius described - starting a X shard node, overloading...
> > The attempt to reproduce the issue was as @avelanarius described above: > > ``` > > * Start a 4-shard single node with some data > > >...
The difference is as follows. Suppose we have table in a keyspace with RF=3. A INSERT query is executed. 3 nodes can become coordinators: A (shard a), B (shard b),...
> I see. I assumed that the optimization you are describing is already the way the coordinator is chosen. > If it's not the case this is an orthogonal issue...
I'll review it
You should also update the docs page about this change (https://github.com/scylladb/python-driver/blob/master/docs/dates-and-times.rst) and possibly other pages if they mention the old way the driver parsed datetimes.
As for some additional testing, I think you should add an additional small test to https://github.com/scylladb/python-driver/blob/27122038b9477eb30ad43f9afdb9f0d79e25182f/tests/integration/standard/test_types.py.