Sebastian Utz
Sebastian Utz
Unfortunately, my previous outcome isn't correct. We still use the dotted representation of nested object columns, e.g. `my_obj.my_inner_column` for the sql subscript representation of `my_obj[my_inner_column]`, inside our table metadata if...
@autophagy Hey Mika, happy new year as well and thanks for the update!
@proddata Since `to_char(timestamp, text)` was released with 4.4.0, did you validate that Looker works now with CrateDB >= v4.4.0?
@proddata great, thx! 🎉
A quick fix could be to add a timeout or iteration loop limit on the where clause optimizer (pk detection). The other fix is probably related to include the analyze/optimize...
Another interesting use-case for replicating single partitions only is a partly-connected remote cluster scenario, mostly when the disconnect is expected and happens over a longer period. E.g. a vessel traveling...
> I fail to understand how this would handle different schemas for different source tables. Is this planned to be possible? No this is not planned and also not required,...
> How could I from 2 source clusters combine tables into one sink table then? On both sides a partitioned table will be created including e.g. a `site_id` into the...
> Who defines the schema? Schema changes on any side will be a problem, indeed. At least for the first iteration I'd suggest to forbid schema changes on partitioned tables...
We've implemented this by #16799 and choose to stop showing error details once more than `10` errors happen on a shard (avoid too much memory consumption by error handling). The...