chanezon / azure-linux Goto Github PK

Patrick, https://github.com/docker/machine has the command as docker-machine create -d azure --azure-subscription-id="SUB_ID" --azure-subscription-cert="mycert.pem"

but you have it as "machine":

$ machine create -d azure
--azure-subscription-id="9b5910a1-...-8e79d5ea2841"
--azure-subscription-cert="azure-cert.pem"
pat-1

Typo, or was there something else about the way you did it that made it different?

hi

i have pfx, .cer, .jks, how to create .pem, create using java keygen..

and create .pem using ssh-keygen (openssl)

i try to link docker-machine..

mervo@SharkBay:~$ docker-machine create -d azure \

--azure-subscription-id="5d6b8a30-2988-xxx-a39a-5502c8f72220"
--azure-subscription-cert="/home/mervo/merv-azure.pem"
--azure-location="East US"
--azure-size=Small
--azure-username="meruvian"
merv-docker-machine-n
Running pre-create checks...
Error with pre-create check: "Get https://management.core.windows.net/5d6b8a30-2988-4700-a39a-5502c8f72220/services/hostedservices/operations/isavailable/merv-docker-machine-n: local error: internal error"
mervo@SharkBay:~$

can help?

https://github.com/chanezon/azure-linux/blob/master/coreos/cluster/azure-coreos-cluster#L152 contains a 'hot' password.

openssl creates keys with passphrase by default, fleetctl cannot ssh to cluster if passphrase is present. Need to remove it first.

openssl rsa -in ~/.ssh/id_rsa -out ~/.ssh/id_rsa_new

A CoreOS cluster is doomed if more than (n/2)-1 of its nodes are lost, posing issues in connection with Azure VM maintenance. A cluster created with the script is currently not protected against initiated reboots of VMs.

Basically, we need to find a way for the script to set up a cluster so that half the machines are always online and live.

I created https://deis.blob.core.windows.net/releases/deis-v1_1_1.vhd which is a data disk mounted as /var/lib/docker on a CoreOS host that has Deis v1.1.1 images pre-cached in the Docker graph.

If the azure-coreos-cluster script has a flag for, say, --data-disk-url, we could attach this disk on boot instead of creating a blank data disk. If we drop format-ephemeral.service from deis-cloud-init-with-disk.yml (but still mount it with var-lib-docker.mount), new machines coming up should already have Deis v1.1.1 cached locally and wouldn't need to pull it from DockerHub. This should significantly speed up Deis deployments on Azure.

Command used
./azure-coreos-cluster xxxxxxxx
--ssh-cert ~/xxxxxxxx/ssh-cert.cer
--ssh-thumb xxxxxxxx
--subscription xxxxxxxx
--azure-cert ~/xxxxxxxx/azure-cert.pem
--num-nodes 5
--location "xxxxxxxx"
--vm-size Large
--vm-name-prefix xxxxxxxx
--virtual-network-name "xxxxxxxx"
--availability-set xxxxxxxx
--deis
--pip
--data-disk
--blob-container-url https://xxxxxxxx.windows.net/vhds/

ERROR:

{'status': u'Failed', 'error': <azure.servicemanagement.OperationError object at 0x111201290>, 'id': u'', 'http_status_code': u'400'}
BadRequest
{'message': u"The subscription policy limit for resource type 'public IP count' was exceeded. The limit for resource type 'public IP count' is 5 per subscription, the current count is 5, and the requested increment is 1.", 'code': u'BadRequest'}
Failed
create VM tlepro-tle_-0 took: 15s
Traceback (most recent call last):
File "./azure-coreos-cluster", line 216, in
role_size=args.vm_size, network_config=network, data_virtual_hard_disks=data_disks)
File "/Library/Python/2.7/site-packages/azure/servicemanagement/servicemanagementservice.py", line 1342, in add_role
async=True)
File "/Library/Python/2.7/site-packages/azure/servicemanagement/servicemanagementclient.py", line 235, in _perform_post
response = self.perform_post(path, body, x_ms_version)
File "/Library/Python/2.7/site-packages/azure/servicemanagement/servicemanagementclient.py", line 162, in perform_post
response = self._perform_request(request)
File "/Library/Python/2.7/site-packages/azure/servicemanagement/servicemanagementclient.py", line 192, in _perform_request
return _management_error_handler(ex)
File "/Library/Python/2.7/site-packages/azure/servicemanagement/init.py", line 1811, in _management_error_handler
return _general_error_handler(http_error)
File "/Library/Python/2.7/site-packages/azure/init.py", line 924, in _general_error_handler
_ERROR_NOT_FOUND.format(str(http_error)))
azure.WindowsAzureMissingResourceError: Not found (Not Found)