Pending jobs in EC2 instance / GL 1.6.1

User 2263 | 10/9/2015, 6:13:34 PM

Hi Nice People,

I'm trying to run a set of jobs in an EC2 instance. Jobs are validated, a working directory is assigned, and they are submitted. However, they stay in 'pending' with no change for hours. Local, asynchronous jobs, run immediately.

A simplified version of my code looks like:

python import graphlab as gl s3_path = 's3://my-s3path' cluster_config = gl.deploy.Ec2Config( region='us-east-1', instance_type='m3.large') ec2_cluster = gl.deploy.ec2_cluster.create( 'Test-Cluster', s3_path, cluster_config) def add(x, y): return x + y job = gl.deploy.job.create( add, environment=ec2_cluster, x=1, y=2)

the server response is

python [INFO] Validating job. [INFO] Validation complete. Job: 'add-Oct-09-2015-13-34-12' ready for execution. [INFO] Job working directory: s3://my-s3path/this-path [INFO] Submitting job... [INFO] Job submitted successfully. [INFO] Job: 'add-Oct-09-2015-13-34-12' scheduled.

When querying the status:

python job.get_status() u'Pending'

Any clues? Thanks

Running GL Create 1.6.1

Comments

User 1178 | 10/10/2015, 6:43:27 PM

Hi,

Thanks for using GraphLab Create and Dato Distributed! Your feedback is greatly appreciated.

This is most likely an issue in our end where we are not effectively using the system resource. We will investigate internally.

In the mean time, you can be unblocked by switching to a slightly bigger EC2 instance. The suggested one to start with is m3.xlarge.

Thanks!

Ping


User 2263 | 10/15/2015, 4:09:39 PM

Hi Ping,

The suggested m3.xlarge EC2 instance run smoothly.

Thank you very much.