dask-ec2
dask-ec2 copied to clipboard
dask
Start a cluster in EC2 for dask.distributed
Trying to set up dask-ec2 on my cluster, and ran into this issue. Here is the stack trace: DEBUG: Uploading file /var/folders/t0/98v23y113_73j625cw8nlz340000gn/T/tmpftoc8c to /tmp/.__tmp_copy DEBUG: Running command sudo -S bash...
Dask-ec2 looks like a gread project! Currently the initial user experience is pretty rocky unfortunately. Neither the default ami, nor amis for Ubuntu 14.04, 16.04, and 17.04 allow me to...
Originally posted to issue #32, but on closer inspection it looks like the errors are slightly different: ``` Error with request: ERROR: Could not connect to salt server. Try `dask-ec2...
As title. Usually, there are several security groups combine together for a instance.
When a Salt command returns a dictionary like `{u'return': [{}]}`, the `print_state` method converts this to a `None` response, causing the following error: ``` Installing scheduler Traceback (most recent call...
Installing workers Traceback (most recent call last): File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/dask_ec2/cli/main.py", line 23, in start cli(obj={}) File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/click/core.py", line 722, in __call__ return self.main(*args, **kwargs) File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/click/core.py", line 697, in main rv...
I created the AMI by starting a dask-ec2 cluster, logging into the 0th node, installing other software dependencies, and then creating an AMI from this node. Then, when starting a...
Destroy terminates the instances: is there an option to stop them and then restart from them? It can save money and time :).
For the CLI, the current default log level has `dask_ec2` set to `DEBUG` and `paramiko` set to `WARNING`. While keeping the default as is, the addition of the following logging...
The default 500GB is far to large in my opinion and leads to large EBS charges. The new 8GB default should be enough for most workloads an matches what is...
