agent: Add start/stop resource to rust agent

[utopiabound]

This is to suppliment python agent start_target. It does not do any movement if the resources is started, nor does it try to force a start in a particular place.

Related #1464

Signed-off-by: Nathaniel Clark nclark@whamcloud.com

---

This change is 

[mkpankov]

Reviewed 3 of 3 files at r1.
Reviewable status: all files reviewed, 4 unresolved discussions (waiting on @utopiabound)

---

iml-agent/src/action_plugins/high_availability/mod.rs, line 82 at r1 (raw file):

fn resource_status(tree: Element) -> Result<HashMap<String, String>, ImlAgentError> {
    let resources = tree.find("resources").unwrap();
    Ok(resources

Wrapping tens of lines in the returned value reads weird

---

iml-agent/src/action_plugins/high_availability/mod.rs, line 90 at r1 (raw file):

                        "resource" => e.get_attr("role").map(|r| vec![(id, r)]),
                        "clone" => {
                            // "Started" -> "Starting" -> "Stopping" -> "Stopped"

Typo?

I expected Starting to be first

---

iml-agent/src/action_plugins/high_availability/mod.rs, line 95 at r1 (raw file):

                                .max_by(|x, y| {
                                    if x == y {
                                        return Ordering::Equal;

This is too much nesting IMO...

---

iml-agent/src/action_plugins/high_availability/mod.rs, line 98 at r1 (raw file):

                                    }
                                    match (*x, *y) {
                                        (_, "Stopped") => Ordering::Greater,

This should go to impl Ord I think. Complement with impl From<&str> for State for ease of use.

[jgrund]

Format check is failing

[jgrund]

Why does this supplement the python start_target instead of replacing it?

[utopiabound]

Why does this supplement the python start_target instead of replacing it?

Because of all the other work that start_target does (looking for paired ZFS resources, and movement of resources)

[jgrund]

A few comments about Result

[johnsonw]

Looks like there is a failure in the integration tests.

[jgrund]

Nit: this is similar to what I'm doing in terms of reading crm_mon resources. We should consolidate our changes.