Nomad task names with slashes fail

[leetrout]

Nomad version

Nomad v0.3.2

Operating system and Environment details

CentOS Linux release 7.2.1511 (Core)
Linux lees-remote-dev 3.10.0-327.10.1.el7.x86_64 #1 SMP Tue Feb 16 17:03:50 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux

Issue

I created a task with the same name as my image. When using a slash in a task name running the job fails with the error:

2016/04/28 22:06:24 [ERR] client: failed to start task 'amos/build-8' for alloc 'd60c2cd5-5e61-6105-7386-ec23fc266d28': error creating rpc client for executor plugin: plugin exited before we could connect

It wasn't immediately clear that the slash was the problem until I saw the logs. Perhaps it would be good to do some or all of the following:

• support this by running os.MkdirAll
• sanitize the name before using it in a path. e.g. slugify it
• check the config and report an accurate error
• add additional details to the documentation that note the task name is used to create files on disk

I'd be happy to contribute a patch if someone has some direction on the best solution(s).

Reproduction steps

Create a job with a task name with a slash.

Nomad Server logs

2016/04/28 22:06:24 [DEBUG] driver.docker: identified image 127.0.0.1:5000/amos/build-8 as sha256:c73e3f8ce8c8efe2fc48a601d2ce001fbdbc0f898928da1eb3ffeb9579d4bd7e
    2016/04/28 22:06:24 [DEBUG] plugin: starting plugin: /usr/local/bin/nomad []string{"/usr/local/bin/nomad", "executor", "/run/nomad/alloc/d60c2cd5-5e61-6105-7386-ec23fc266d28/amos/build-8/amos/build-8-executor.out"}
    2016/04/28 22:06:24 [DEBUG] plugin: waiting for RPC address for: /usr/local/bin/nomad
    2016/04/28 22:06:24 [DEBUG] plugin: nomad: open /run/nomad/alloc/d60c2cd5-5e61-6105-7386-ec23fc266d28/amos/build-8/amos/build-8-executor.out: no such file or directory
    2016/04/28 22:06:24 [DEBUG] plugin: /usr/local/bin/nomad: plugin process exited

Job file

Bad job

job "example" {
        datacenters = ["dc1"]
        type = "batch"
        group "amos-b-8" {
                task "amos/build-8" {
                        driver = "docker"
                        config {
                                image = "127.0.0.1:5000/amos/build-8"
                        }
                        resources {
                                cpu = 500 # 500 Mhz
                                memory = 500 # 256MB
                                network {
                                        mbits = 10
                                }
                        }
                }
        }
}

Good job

job "example" {
        datacenters = ["dc1"]
        type = "batch"
        group "amos-b-8" {
                task "amos-build-8" {
                        driver = "docker"
                        config {
                                image = "127.0.0.1:5000/amos/build-8"
                        }
                        resources {
                                cpu = 500 # 500 Mhz
                                memory = 500 # 256MB
                                network {
                                        mbits = 10
                                }
                        }
                }
        }
}

[dadgar]

Hey @leetrout,

I think our best path forward is disallowing tasks containing the slash from being submitted. This can be accomplished by patching the validation logic.

https://github.com/hashicorp/nomad/blob/master/nomad/structs/structs.go#L1708

Look forward to your PR 👍

[leetrout]

Great, thanks for the direction.

[github-actions]

I'm going to lock this issue because it has been closed for 120 days ⏳. This helps our maintainers find and focus on the active issues.
If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.