Creating hosts from Rancher on DigitalOcean fails with various error messages

Hi,

At DO Support’s recommendation set up some infrastructure with Rancher last week. Rancher is on a 1GB instance and I was able to bring up multiple hosts and set up LB. Very slick.

Went to demo it to the team this morning and I can’t create hosts and I get back a variety of vague error messages. This morning it started with “Exit code 2”. Later I got “Exit code 1”. Just tried again and got “fork/exec /usr/bin/docker-machine: cannot allocate memory”. These failures occur right away. The API view from my last attempt follows…

{
"id": "1ph24",
"type": "machine",
"links": {
"self": "…/v1/projects/1a83/machines/1ph24",
"account": "…/v1/projects/1a83/machines/1ph24/account",
"hosts": "…/v1/projects/1a83/machines/1ph24/hosts",
},
"actions": {
"remove": "…/v1/projects/1a83/machines/1ph24/?action=remove",
},
"name": "ssx",
"state": "error",
"accountId": "1a83",
"amazonec2Config": null,
"authCertificateAuthority": null,
"authKey": null,
"azureConfig": null,
"created": "2016-01-04T21:37:30Z",
"createdTS": 1451943450000,
"description": "A test host",
"digitaloceanConfig": {
"accessToken": "sanitized for our protection",
"size": "1gb",
"region": "nyc3",
"image": "ubuntu-14-04-x64",
"privateNetworking": true,
"sshUser": null,
},
"dockerVersion": null,
"driver": "digitalocean",
"engineEnv": null,
"engineInsecureRegistry": null,
"engineInstallUrl": null,
"engineLabel": null,
"engineOpt": null,
"engineRegistryMirror": null,
"engineStorageDriver": null,
"exoscaleConfig": null,
"externalId": "c08a1038-9835-43f2-850c-2eb56c0fd88d",
"kind": "machine",
"labels": {
"role": "foo",
},
"openstackConfig": null,
"packetConfig": null,
"rackspaceConfig": null,
"removed": null,
"softlayerConfig": null,
"transitioning": "error",
"transitioningMessage": "fork/exec /usr/bin/docker-machine: cannot allocate memory",
"transitioningProgress": null,
"ubiquityConfig": null,
"uuid": "3f5db28c-5f65-45c4-bf5a-bc4b610e2e87",
"virtualboxConfig": null,
"vmwarevcloudairConfig": null,
"vmwarevsphereConfig": null,
}

I can create a droplet (host) on DigitalOcean’s page without issue.

Any ideas where to begin?

Are you creating hosts using the UI through Rancher? Typically, the error messages come from docker-machine as we are just running a docker-machine command.

I happen to notice that the privateNetworking is set to true, which I don’t usually play with. Can you share any other options you have set in the UI?

Thanks, Denise.

I suspect DigitalOcean was having issues. Later on they blocked all droplet creates. I’ve since dried up all my droplets and started over and everything is working as expected.

Given that Rancher is a cowboy herding the cattle (see what I did there? :slightly_smiling: ) with the cattle being docker-machine and friends, it would probably be good to always preface every error message with something like “docker-machine on 192.168.1.1 reports ‘exit status 0’”. You folks intuitively know from where error messages may originate because you live Rancher. The rest of us are probably just trying to get work done.

Thanks!

I like that idea of the “docker-machine reports:” for those messages.

I’ve created an enhancement request.