tkram01
tkram01
I think I am seeing the same issue. This code runs sequentially ``` from time import sleep from prefect import Flow, task from prefect.executors import LocalDaskExecutor @task() def add_ten(x): print(f'starting...
+1 especially image
We would like the option to use pickle as well. Some swig wrappers support pickle serialization and not JSON.
I am getting a similar error. I am using the Ray AWS cluster launcher. When I run a raydp program it acts like it is using a local ray instance...
@ConeyLiu I get the similar behavior with your code. With the increased value for executor_memory I get these messages when monitoring. ```======== Autoscaler status: 2021-02-22 01:11:35.636059 ======== Node status ---------------------------------------------------------------...
@ConeyLiu I see similar behavior on 1.1.0 ```2021-02-22 13:27:35,625 INFO resource_demand_scheduler.py:148 -- Cluster resources: [{'memory': 20.0, 'node:172.31.12.88': 1.0, 'CPU': 2.0, 'object_store_memory': 7.0}] 2021-02-22 13:27:35,625 INFO resource_demand_scheduler.py:149 -- Node counts: defaultdict(,...
Yes - I'm happy to work on it with some guidance on the merge process. Thanks!
having the same issue
Exposing IGRAPH_ATTRIBUTE_OBJECT to the C attribute handler sounds like a great solution. We have some complicated objects that we want to attach to each node that we use in isomorphism...
We would be really interested in this feature.