Setup problems: AWS

@rachel Hi Rachel, I was able to get T2 to run last week and get into the notebooks. But when I try to get into it again today I keep hitting the limit errors.

aws: error: argument --group-id: expected one argument
setup_p2.sh: line 13: /home/vagrant/.ssh/aws-key.pem: Permission denied
usage: aws [options] [ …] [parameters]
To see help text, you can run:

aws help
aws help
aws help
aws: error: argument --subnet-id: expected one argument

An error occurred (AddressLimitExceeded) when calling the AllocateAddress operation: The maximum number of addresses has been reached.
Waiting for instance start…

I have gotten approval from AWS but it was after I created my user, do I need to do anything to switch it from t2 to p2?

After some research, I also followed instructions to Change EC2 Size. To aws ec2 modify-instance-attribute --instance-id $instanceId --instance-type t2.large
But I still have the limit error even after that.

This might also be helpful: When I checked to see if I have an instanceid for p2 or t2, I got an instance for t2 but not p2.

(env)vagrant@vagrant-ubuntu-trusty-64:~/src/deep-learning$ aws-get-p2
None
(env)vagrant@vagrant-ubuntu-trusty-64:~/src/deep-learning$ aws-get-t2
i-090
(env)vagrant@vagrant-ubuntu-trusty-64:~/src/deep-learning$ aws-start i-090
STARTINGINSTANCES i-090
CURRENTSTATE 0 pending
PREVIOUSSTATE 80 stopped
35.161.163.63 i-090

Unsure what my next steps should be. Thank you in advance!

@lingsitu1290 @zoha.zargham please complete the steps that @rachel provided last night in this thread: http://wiki.fast.ai/index.php/Starting_Over_with_AWS . Then try again.

@jeremy
Thanks .
I did the steps.
I get error on bash setup_p2.sh
An error occurred (InvalidKeyPair.Duplicate) when calling the CreateKeyPair operation: The keypair ‘aws-key’ already exists.

@zoha.zargham that means you missed the step that deletes all your existing keys. Please try going through the steps again, and note that step in particular

@jeremy

Got through the steps too and ran setup_p2.sh.

Got:

An error occurred (InvalidKeyPair.NotFound) when calling the RunInstances operation: The key pair ‘aws-key’ does not exist
Waiting for instance start…

@jeremy Please disregard, figured it out.

@rachel Followed the directions for starting over, works perfectly fine now. Thank you :slight_smile:

1 Like

@lingsitu1290 @rachel
hey, I get the same error. How did you resolve?

@zoha.zargham
Have to run:
rm ~/.ssh/aws-key.pem

before bash setup_p2.sh

it worked after!

@zoha.zargham The common problem section of the AWS Install wiki is also a helpful resource for AWS errors: http://wiki.fast.ai/index.php/AWS_install#Common_Problems

@nichol I am connecting to aws via cygwin on Windows 7. I was fighting with the cygwin and windows conflicts, but below are some of the things I had to did to resolve them.

I follow instructions in this video (by Jeremy) https://www.youtube.com/watch?v=UV6WIBi9GJQ

Install cygwin (http://www.cygwin.com/):
Explicitly select the following to be installed and not skipped:

  • wget
  • python (2.7 only)
  • openssh (this was required else cygwin was using my Windows SSH installed)

Install pip on Cygwin:
Run from cygwin:
$wget https://bootstrap.pypa.io/get-pip.py
$python get-pip.py

I have a rather silly question. After we run our setup script, created an instance. How come there is a folder nbs already there and we all have the same password dl_course… ? Is it because it’s all written in the setup script?

I believe that is because it came from the AMI which in turn came from the install_gpu script.

2 Likes

I am looking directly for the AMI on AWS and I can’t see it - also had problems related to the command line for the same reason.

Does the AMI still exist?

trying to install aws in cygwin, but apparently the there is already awscli installed within my anaconda and hence when I typed aws in cygwin I got this error message: C:\users\fgoza\anaconda3\python.exe: can’t open file ‘/cygdrive/c/Users/fgoza/Anaconda3/Scripts/aws’: [Errno 2] No such file or directory

have tried this: http://wiki.fast.ai/index.php/Awscli_in_cygwin
with no avail, it is still installed into anaconda

suggestions please ?

sml0820 I had this problem too. The AMI only exists on us-west-2, you are on us-east-1 ( the N. Virginia)

1 Like

I probably took the long way around here but I installed the ubuntu bash shell: http://www.howtogeek.com/249966/how-to-install-and-use-the-linux-bash-shell-on-windows-10/ as @jeremy suggested to @nichol.bradford a few weeks back. When I still had trouble with pip install blowing-up I switch to sudo apt-get install which Jeremy also suggested but I missed the first time – life has never been better. Now, onto Identity and Access Management which by the way looks like it’s been updated since ~October.

Hope this helps!

1 Like

Is there a way to delete your default VPCs? I keep getting an error that the maximum number of VPCs has been reached when running the setup_t2.sh script. I tried searching online but I have yet to find a viable solution.

Steve, your suggestion worked for me. I have no experience with Linux and it took an entire morning fiddling around with this to get the configuration to work on Windows 10 box. I tried installing Cygwin and executing the setup_t2.sh script within bash as indicated in the setup video, but; Cygwin did not find find the aws script. The Linux bash shell on windows however worked like a charm. I also had to manually fix the .ssh key directory. Thanks again!

1 Like

The wiki FAQ is your friend :slight_smile: http://wiki.fast.ai/index.php/Frequently_Asked_Questions

But you shouldn’t run setup_t2.sh more than once. Why are you running it multiple times?

1 Like