CLI Allow querying all namespaces for jobs and allocations - Try 2

[notnoop]

This is attempt 2 of #8183 .

This is almost the same approach, but reuses current namespace field but with specially handling * to mean all namespaces.

UX

When -namespace=* command line argument is set, or if NOMAD_NAMESPACES env var is set to *, nomad job and nomad alloc commands will lookup relevant jobs/allocations in all namespaces they are authorized to.

nomad job status -namespace=* will emit all the jobs across namespaces (with a namespace column added). nomad job status -namespace=* foo will return the job info for foo job regardless of which namespace is in; if there multiple foo jobs, the CLI will emit back an error message, indicating the possible values.

# namespace specific status
$ nomad job status
ID        Type   Priority  Status   Submit Date
example1  batch  50        running  2020-06-17T10:49:26-04:00

# list jobs across all namespaces
$ nomad job status -namespace=*
ID        Namespace  Type   Priority  Status   Submit Date
example1  default    batch  50        running  2020-06-17T10:49:26-04:00
example1  qa         batch  50        running  2020-06-17T10:49:37-04:00
qa-job    qa         batch  50        running  2020-06-17T10:50:28-04:00

# get job status for a job in a specific namespace
$ nomad job status -short ex
ID            = example1
Name          = example1
Submit Date   = 2020-06-17T10:49:26-04:00
Type          = batch
Priority      = 50
Datacenters   = dc1
Namespace     = default
Status        = running
Periodic      = false
Parameterized = false

# when checking multiple namespaces, report an error message if multiples are found
$ nomad job status -short -namespace=* ex
Prefix matched multiple jobs

ID        Namespace  Type   Priority  Status   Submit Date
example1  default    batch  50        running  2020-06-17T10:49:26-04:00
example1  qa         batch  50        running  2020-06-17T10:49:37-04:00

# but namespace can always be specified to disambiguate
$ nomad job status -short -namespace qa ex
ID            = example1
Name          = example1
Submit Date   = 2020-06-17T10:49:37-04:00
Type          = batch
Priority      = 50
Datacenters   = dc1
Namespace     = qa
Status        = running
Periodic      = false
Parameterized = false

# querying with a prefix if a uniue job is found works
$ nomad job status -short -namespace=* qa
ID            = qa-job
Name          = qa-job
Submit Date   = 2020-06-17T10:50:28-04:00
Type          = batch
Priority      = 50
Datacenters   = dc1
Namespace     = qa
Status        = running
Periodic      = false
Parameterized = false

Implementation

The implementation is very identical to other approach

Trade-offs?

The advantage of this approach is that it's simpler to explain, has fewer moving parts, and provides a simple generalizable approach for other fields.

The downside though is that operators may need to fine tune namespace parameter in invocation basis. If a user sets NOMAD_NAMESPACE=*, they may get odd errors if hitting endpoints that don't support this yet (e.g. evals). Though, we should try to extend them to eliminate this annoying inconsistency if they exist.

[schmichael]

Seems like we could add globbing pretty easily since:

1. we already use https://github.com/ryanuber/go-glob
2. we already loop over every namespace to build the set
3. we already loop over all objects in every namespace

So it'd be a little more CPU work per-namespace, but not listed per-object, so it seems fine.

Not a blocker for this.

[github-actions]

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