Re: StarCluster Digest, Vol 66, Issue 7
Cannot thank you enough for detailed handling my novice queries. Very kind
of you. Your VPC configuration, AMI and volume mounting suggestions have
made life worth now ;) . Thanks a lot.
On 23 February 2015 at 23:18, Jennifer Staab <jstaab_at_cs.unc.edu> wrote:
> Seems like you needed to alter your custom VPC so that it allows access to
> an internet gateway and/or there is a public IP/ or Elastic IP associated
> to instance that internet traffic is being routed through (like a NAT). If
> you aren't using a NAT server to route traffic through, your Master node
> needs to be associated with a subnet that has a route table that allows
> access to the internet gateway and the instance needs to be assigned a
> Public IP or Elastic IP address. Amazon created the default VPC's so that
> the typical user could launch their instances in the VPC and use them
> without having to know all the details of setting up subnets, route tables,
> NACLs and the like.
>
> From the account you deleted the default VPC, you still have access to
> other default VPCs in different regions as long as you didn't delete those
> also. You could look at their setup to help you fix your custom VPC to
> better match the settings of the default VPC that you deleted. Regions are
> geographic and don't share resources. So if you want a resource (AMI, EBS
> volume), available in a different region (in the same account) -- locate or
> take a recent "snapshot" of it. From the AWS console, go to EC2 Services
> then select "Snapshots" and find the snapshot you want to make available to
> a different region and select it. Then select "Actions" and select "Copy"
> -- from there you can select your destination region so it that snapshot
> will be available in a different region.
>
> As for making an AMI or Snapshot available to a different account you have
> to login to AWS console, get to EC2 Services, select "Snapshots" and then
> select snapshot you want to make available to another account, then select
> "Actions" and select "Modify Snapshot Permissions" and add permission to
> your other AWS account and click "save". For AMI's it's a similar process,
> select "AMIs", select the AMI you want to make available, and select
> "Modify Image Permissions" etc.
>
> It is a little unclear to me exactly what issue you are having from what
> you described in your email, there are many ways things could have gone
> wrong. First thing is check that the EBS volume you attached to your master
> node is actually fully attached to the volume from the AWS console -
> specifically its State "in-use" and Attachment Information is "attached".
> The instance you want to attach to AND your EBS volume need to be in the
> same availability zone for you to make the attachment. After you are
> certain it is attached via the AWS console, login to the master node and
> run the command "lsblk" if the volume is truly attached you should see it
> in the list (if you attached volume as /dev/sdp -- you should see it as
> /dev/xvdp). Next you need to mount the volume with command "sudo mount
> <device name> <volume name>" where device name is how you attached the
> volume and volume name is the mount point you are using. Lets say I
> attached my volume as /dev/sdp and my my mount point is /mntPt. So I would
> issue command "sudo mount /dev/xvdp /mntPt" to mount the device. For my
> example, if you do ls -lh /mntPt you should see all information on mounted
> volume. If you want to make the EBS volume available via NFS to all nodes
> in your cluster, there is more that needs to be done. Easiest way to do
> this is to use Starcluster config and configure the EBS volume to be used
> with the Starcluster Cluster (see here
> <http://star.mit.edu/cluster/docs/latest/manual/configuration.html#amazon-ebs-volumes>
> for details). Other issues you might have are if you created volume from
> scratch and you need to format the volume to have a file system prior to
> mounting it. Also on occasion, I have had issue mounting SSD volumes if
> there is a partition involved (e.g. use /dev/xvdp1 instead of /dev/xvdp as
> device name) .
>
> Good Luck,
> -Jennifer
>
>
Received on Tue Feb 24 2015 - 04:51:04 EST
This archive was generated by
hypermail 2.3.0.