Closed delgadom closed 5 years ago
Scaling using the widget has no effect, and calling cluster.scale(n) # n < cluster size, including 0 results in the following:
cluster.scale(n) # n < cluster size, including 0
--------------------------------------------------------------------------- AttributeError Traceback (most recent call last) <ipython-input-48-63c9907b662c> in <module>() ----> 1 cluster.scale(0) /opt/conda/lib/python3.6/site-packages/dask_kubernetes/core.py in scale(self, n) 371 return 372 --> 373 to_close = select_workers_to_close(self.scheduler, n_to_delete) 374 logger.debug("Closing workers: %s", to_close) 375 if len(to_close) < len(self.scheduler.workers): /opt/conda/lib/python3.6/site-packages/dask_kubernetes/core.py in select_workers_to_close(scheduler, n_to_close) 536 assert n_to_close <= len(workers) 537 key = lambda ws: ws.info['memory'] --> 538 to_close = set(sorted(scheduler.idle, key=key)[:n_to_close]) 539 540 if len(to_close) < n_to_close: /opt/conda/lib/python3.6/site-packages/dask_kubernetes/core.py in <lambda>(ws) 535 workers = list(scheduler.workers.values()) 536 assert n_to_close <= len(workers) --> 537 key = lambda ws: ws.info['memory'] 538 to_close = set(sorted(scheduler.idle, key=key)[:n_to_close]) 539 AttributeError: info
This occurs on v0.2.4 images, and appears to be fixed in dask/dask-kubernetes#104
closed by #73
Scaling using the widget has no effect, and calling
cluster.scale(n) # n < cluster size, including 0
results in the following:This occurs on v0.2.4 images, and appears to be fixed in dask/dask-kubernetes#104